headings; a heading","icon":"$undefined","url":"/docs/usage/run-recipes/backlog-tasks--whats-pending-beyond-run-memory","$ref":"usage/run-recipes/backlog-tasks--whats-pending-beyond-run-memory.md"},{"$id":"usage/run-recipes/claude-code-memory-adapter-cyberine_cc_memory--opt-in.md","type":"page","name":"Claude Code memory adapter (CYBERINE_CC_MEMORY) — opt-in","description":"Cyberine's memory is its own: ~/.cyberine/memory/ (the Long-term memory section above), one format, vendor-independent, so the same orchestrator serves Claude, Codex or any other CLI. A Claude...","icon":"$undefined","url":"/docs/usage/run-recipes/claude-code-memory-adapter-cyberine_cc_memory--opt-in","$ref":"usage/run-recipes/claude-code-memory-adapter-cyberine_cc_memory--opt-in.md"},{"$id":"usage/run-recipes/shared-memory-backend--contextq-or-any-similar-server.md","type":"page","name":"Shared memory backend — ContextQ (or any similar server)","description":"The local episodic store above is the canonical record store and is always on. An OPTIONAL augmentation backend, selected purely by config, ALSO receives each terminal run and serves a...","icon":"$undefined","url":"/docs/usage/run-recipes/shared-memory-backend--contextq-or-any-similar-server","$ref":"usage/run-recipes/shared-memory-backend--contextq-or-any-similar-server.md"},{"$id":"usage/run-recipes/contextbrain--code-context-3-act-adapter-localbrain--contextqbrain-t441t454.md","type":"page","name":"ContextBrain — Code-Context 3-Act adapter (LocalBrain / ContextQBrain, T441/T454)","description":"Separate from the memory-preamble backend above, the ContextBrain port (src/core/contextBrain.ts) is the seam behind the Code-Context 3-Act architecture (docs/design/code-context-3act.md) used by","icon":"$undefined","url":"/docs/usage/run-recipes/contextbrain--code-context-3-act-adapter-localbrain--contextqbrain-t441t454","$ref":"usage/run-recipes/contextbrain--code-context-3-act-adapter-localbrain--contextqbrain-t441t454.md"},{"$id":"usage/run-recipes/project-groups--link-dirs-to-share-runs--chats--memory.md","type":"page","name":"Project groups — link dirs to share runs / chats / memory","description":"A GROUP links several project directories so their runs, chats, and long-term memory collaborate across projects (one Cyberine group = one shared memory pool). Groups are OPT-IN: with no config,...","icon":"$undefined","url":"/docs/usage/run-recipes/project-groups--link-dirs-to-share-runs--chats--memory","$ref":"usage/run-recipes/project-groups--link-dirs-to-share-runs--chats--memory.md"},{"$id":"usage/run-recipes/knowledge-ledger--cyberine-knowledge-k6tbw.md","type":"page","name":"Knowledge ledger — cyberine knowledge (k6tbw)","description":"An append-only ledger of lessons, facts, pointers and prefs that lives outside every repo, under $CYBERINEHOME/knowledge/ (default ~/.cyberine/knowledge/). Nothing is written inside the repo...","icon":"$undefined","url":"/docs/usage/run-recipes/knowledge-ledger--cyberine-knowledge-k6tbw","$ref":"usage/run-recipes/knowledge-ledger--cyberine-knowledge-k6tbw.md"},{"$id":"usage/run-recipes/expose-cyberine-as-an-mcp-server--npm-run-mcp.md","type":"page","name":"Expose Cyberine as an MCP server — npm run mcp","description":"npm run mcp starts Cyberine as an MCP server (stdio transport) so any MCP host (Claude Code, Cursor, Claude Desktop) can drive it. The server declares a short instructions string (a few lines: what","icon":"$undefined","url":"/docs/usage/run-recipes/expose-cyberine-as-an-mcp-server--npm-run-mcp","$ref":"usage/run-recipes/expose-cyberine-as-an-mcp-server--npm-run-mcp.md"},{"$id":"usage/run-recipes/consume-external-mcp-servers-as-tools--mcptooladapter.md","type":"page","name":"Consume external MCP servers as tools — McpToolAdapter","description":"Cyberine can consume any external MCP server's tools as orchestrator Tools (the consume-side mirror of npm run mcp, which exposes Cyberine itself). Configure servers by env; nothing connects when...","icon":"$undefined","url":"/docs/usage/run-recipes/consume-external-mcp-servers-as-tools--mcptooladapter","$ref":"usage/run-recipes/consume-external-mcp-servers-as-tools--mcptooladapter.md"},{"$id":"usage/run-recipes/oversized-tool-results--spill-to-disk--retrieval-spill_retrieve.md","type":"page","name":"Oversized tool results — spill-to-disk + retrieval (spill_retrieve)","description":"Every tool call an agent adapter makes (LoopAgentAdapter, AnthropicAgentAdapter, BedrockAdapter, VertexAdapter — including MCP tools consumed via McpToolAdapter, which previously had no cap at...","icon":"$undefined","url":"/docs/usage/run-recipes/oversized-tool-results--spill-to-disk--retrieval-spill_retrieve","$ref":"usage/run-recipes/oversized-tool-results--spill-to-disk--retrieval-spill_retrieve.md"},{"$id":"usage/run-recipes/history-compaction-spills-oversized-tool-results-instead-of-truncating-to-80-chars.md","type":"page","name":"History compaction spills oversized tool results instead of truncating to 80 chars","description":"For a worker with historyCompression: true on its WorkerProfile (see src/core/workerProfiles.ts), once accumulated message history exceeds historyCompressionBudget (default 8000 tokens,...","icon":"$undefined","url":"/docs/usage/run-recipes/history-compaction-spills-oversized-tool-results-instead-of-truncating-to-80-chars","$ref":"usage/run-recipes/history-compaction-spills-oversized-tool-results-instead-of-truncating-to-80-chars.md"},{"$id":"usage/run-recipes/malformed-tool-calls--validate-then-repair-on-the-loop-tool-input-path-kmk7x.md","type":"page","name":"Malformed tool calls — validate-then-repair on the loop tool-input path (kmk7x)","description":"Open models on OpenAI-compatible endpoints (DeepSeek, MiniMax, GLM, Qwen) emit a small set of recurring tool-call shape errors. Before this change a loop worker parsed the arguments with...","icon":"$undefined","url":"/docs/usage/run-recipes/malformed-tool-calls--validate-then-repair-on-the-loop-tool-input-path-kmk7x","$ref":"usage/run-recipes/malformed-tool-calls--validate-then-repair-on-the-loop-tool-input-path-kmk7x.md"},{"$id":"usage/run-recipes/edit-pacing--a-loop-implementer-writes-before-its-round-budget-runs-out.md","type":"page","name":"Edit pacing — a loop implementer writes before its round budget runs out","description":"A key-only loop implementer (loop-deepseek-build and the other loop--build workers) used to spend every tool round reading and never call a write tool. When the budget ran out, the loop asked for...","icon":"$undefined","url":"/docs/usage/run-recipes/edit-pacing--a-loop-implementer-writes-before-its-round-budget-runs-out","$ref":"usage/run-recipes/edit-pacing--a-loop-implementer-writes-before-its-round-budget-runs-out.md"},{"$id":"usage/run-recipes/repeat-tool-nudge--advisory-reminder-when-a-worker-repeats-the-exact-same-call.md","type":"page","name":"Repeat-tool nudge — advisory reminder when a worker repeats the exact same call","description":"safetyRails.checkPreCallCaps (iteration + token caps) runs once PER DISPATCH; a worker grepping the same pattern forty times inside ONE dispatch is invisible to it, and the only thing that...","icon":"$undefined","url":"/docs/usage/run-recipes/repeat-tool-nudge--advisory-reminder-when-a-worker-repeats-the-exact-same-call","$ref":"usage/run-recipes/repeat-tool-nudge--advisory-reminder-when-a-worker-repeats-the-exact-same-call.md"},{"$id":"usage/run-recipes/durable-worker-transcripts--cyberine-transcript.md","type":"page","name":"Durable worker transcripts — cyberine transcript","description":"A loop worker's events.jsonl is telemetry, not a transcript: each entry's detail is a 120-character slice of the prompt, so when a fleet worker fails there was previously no way to see what it was...","icon":"$undefined","url":"/docs/usage/run-recipes/durable-worker-transcripts--cyberine-transcript","$ref":"usage/run-recipes/durable-worker-transcripts--cyberine-transcript.md"},{"$id":"usage/run-recipes/capability-aware-routing--cyberine-route.md","type":"page","name":"Capability-aware routing — cyberine route","description":"Cyberine routes each task by marginal cost so it works standalone (a Claude Code / Codex replacement) on whatever you already pay for. Every worker carries a costModel and a priorityRank...","icon":"$undefined","url":"/docs/usage/run-recipes/capability-aware-routing--cyberine-route","$ref":"usage/run-recipes/capability-aware-routing--cyberine-route.md"},{"$id":"usage/run-recipes/smart-llm-as-router-for-low-confidence-tasks-t61.md","type":"page","name":"Smart (LLM-as-router) for low-confidence tasks (T61)","description":"When the deterministic router is unsure about a task (confidence < 0.7 — the \"default fallback\" bucket), you can opt in to a secondary LLM call that refines the worker ordering and per-worker...","icon":"$undefined","url":"/docs/usage/run-recipes/smart-llm-as-router-for-low-confidence-tasks-t61","$ref":"usage/run-recipes/smart-llm-as-router-for-low-confidence-tasks-t61.md"},{"$id":"usage/run-recipes/router-eval--compare-routing-strategies-on-the-corpus-t66.md","type":"page","name":"Router eval — compare routing strategies on the corpus (T66)","description":"Replays .local/runs/ offline and compares three routing strategies against each step's recorded judge verdict:","icon":"$undefined","url":"/docs/usage/run-recipes/router-eval--compare-routing-strategies-on-the-corpus-t66","$ref":"usage/run-recipes/router-eval--compare-routing-strategies-on-the-corpus-t66.md"},{"$id":"usage/run-recipes/model-strength-table-and-fleetjson-override-kyhae.md","type":"page","name":"Model strength table and fleet.json override (kyhae)","description":"Cyberine ships a sourced answer to \"which model is strong and which is weak\": every model id the fleet can run has a coarse rank from 1 to 5 (5 = strongest) and a cost class from 1 to 3, keyed by...","icon":"$undefined","url":"/docs/usage/run-recipes/model-strength-table-and-fleetjson-override-kyhae","$ref":"usage/run-recipes/model-strength-table-and-fleetjson-override-kyhae.md"},{"$id":"usage/run-recipes/fleet-bench--solve-rate-and-dollars-per-task-kvxah.md","type":"page","name":"Fleet bench — solve rate and dollars per task (kvxah)","description":"cyberine eval bench scores every available worker, and the judged winner, on a versioned suite of coding tasks with programmatic pass/fail. Router eval above compares routing decisions against...","icon":"$undefined","url":"/docs/usage/run-recipes/fleet-bench--solve-rate-and-dollars-per-task-kvxah","$ref":"usage/run-recipes/fleet-bench--solve-rate-and-dollars-per-task-kvxah.md"},{"$id":"usage/run-recipes/history-replay-bench-k0t77.md","type":"page","name":"History replay bench (k0t77)","description":"cyberine eval replay discovers past repository commits that added a scripts/smoke.ts grader and changed src/, starting on 2026-08-01. Each task starts from a git archive export of the parent...","icon":"$undefined","url":"/docs/usage/run-recipes/history-replay-bench-k0t77","$ref":"usage/run-recipes/history-replay-bench-k0t77.md"},{"$id":"usage/run-recipes/chat-prompt-cache-bench-kkrsb.md","type":"page","name":"Chat prompt-cache bench (kkrsb)","description":"npm run bench:chat-cache replays scripted chat turns through the real plannerTurn loop (real system prompt, real tool list, real read-only tools) and reads the provider's own cached-token count...","icon":"$undefined","url":"/docs/usage/run-recipes/chat-prompt-cache-bench-kkrsb","$ref":"usage/run-recipes/chat-prompt-cache-bench-kkrsb.md"},{"$id":"usage/run-recipes/llm-candidate-re-ranker--act-1-scope-filter-t446.md","type":"page","name":"LLM candidate re-ranker — Act 1 SCOPE filter (T446)","description":"An optional one-call sonnet-tier filter that runs during LocalBrain.boot() (Act 1 SCOPE). After the task-text extractor and git co-change ranker assemble a combined ~15-file candidate set, the...","icon":"$undefined","url":"/docs/usage/run-recipes/llm-candidate-re-ranker--act-1-scope-filter-t446","$ref":"usage/run-recipes/llm-candidate-re-ranker--act-1-scope-filter-t446.md"},{"$id":"usage/run-recipes/per-task-model-variant-selection-t60.md","type":"page","name":"Per-task model variant selection (T60)","description":"Loop workers that expose a variants map in their WorkerProfile run a different model id depending on task tier — the router picks the cheapest capable variant so you don't pay for deep reasoning...","icon":"$undefined","url":"/docs/usage/run-recipes/per-task-model-variant-selection-t60","$ref":"usage/run-recipes/per-task-model-variant-selection-t60.md"},{"$id":"usage/run-recipes/machine-readable-route-recommendation--cyberine-route---json-t62.md","type":"page","name":"Machine-readable route recommendation — cyberine route --json (T62)","description":"cyberine route --json \"<task>\" emits a stable JSON object designed for non-Cyberine callers (e.g. a Claude Code subagent dispatcher) that need to ask Cyberine \"what tier should this task run at\"...","icon":"$undefined","url":"/docs/usage/run-recipes/machine-readable-route-recommendation--cyberine-route---json-t62","$ref":"usage/run-recipes/machine-readable-route-recommendation--cyberine-route---json-t62.md"},{"$id":"usage/run-recipes/routing-explain--why-a-worker-was-picked-or-skipped-kc0nz.md","type":"page","name":"Routing-explain — why a worker was picked or skipped (kc0nz)","description":"Routing used to be a black box once the decision was made: reasons[] explained the winning fleet, but nothing recorded which OTHER candidates existed, or why a worker never even reached the router...","icon":"$undefined","url":"/docs/usage/run-recipes/routing-explain--why-a-worker-was-picked-or-skipped-kc0nz","$ref":"usage/run-recipes/routing-explain--why-a-worker-was-picked-or-skipped-kc0nz.md"},{"$id":"usage/run-recipes/opentelemetry-genai-export--cyberine_otel_endpoint-kyv0t.md","type":"page","name":"OpenTelemetry GenAI export — CYBERINE_OTEL_ENDPOINT (kyv0t)","description":"run.json/events.jsonl is a proprietary, Cyberine-only format with no interop. Setting CYBERINEOTELENDPOINT turns on an exporter that maps RunStore run/step/worker events onto spans and ships them...","icon":"$undefined","url":"/docs/usage/run-recipes/opentelemetry-genai-export--cyberine_otel_endpoint-kyv0t","$ref":"usage/run-recipes/opentelemetry-genai-export--cyberine_otel_endpoint-kyv0t.md"},{"$id":"usage/run-recipes/swarm-mode--cyberine-run---mode-swarm-plan---parallel-chunk-implement---merge.md","type":"page","name":"Swarm mode — cyberine run --mode swarm (plan -> parallel-chunk implement -> merge)","description":"For tasks that decompose cleanly into disjoint chunks — large multi-file refactors, sweep updates across many files, any scope where one worker would burn tool rounds but N workers can split the...","icon":"$undefined","url":"/docs/usage/run-recipes/swarm-mode--cyberine-run---mode-swarm-plan---parallel-chunk-implement---merge","$ref":"usage/run-recipes/swarm-mode--cyberine-run---mode-swarm-plan---parallel-chunk-implement---merge.md"},{"$id":"usage/run-recipes/lane-grouping--when-lanes-help-vs-flat-fan-out.md","type":"page","name":"Lane grouping — when lanes help vs flat fan-out","description":"The T269 lane runner is the default swarm executor in --mode swarm. It runs the plan through clusterChunks + packLanes BEFORE workers fan out, so the chunks a worker can see on disk follow the...","icon":"$undefined","url":"/docs/usage/run-recipes/lane-grouping--when-lanes-help-vs-flat-fan-out","$ref":"usage/run-recipes/lane-grouping--when-lanes-help-vs-flat-fan-out.md"},{"$id":"usage/run-recipes/operating-modes----mode-competeswarmdeepsmartrushharvest--cyberine_mode-env.md","type":"page","name":"Operating modes — --mode compete|swarm|deep|smart|rush|harvest (+ CYBERINE_MODE env)","description":"Six preset bundles (PRESETS in src/core/modePresets.ts) that drive WorkflowEngine behaviour end-to-end: fleet size cap, judge chain selection, verify gate, governor, and M5 escalation. Selected...","icon":"$undefined","url":"/docs/usage/run-recipes/operating-modes----mode-competeswarmdeepsmartrushharvest--cyberine_mode-env","$ref":"usage/run-recipes/operating-modes----mode-competeswarmdeepsmartrushharvest--cyberine_mode-env.md"},{"$id":"usage/run-recipes/research-mode--cyberine-research-plan---search---verify---synthesize.md","type":"page","name":"Research mode — cyberine research (plan -> search -> verify -> synthesize)","description":"For questions that need current web evidence — comparisons, market scans, \"what is the state of X in 2026?\", trade-off write-ups — use cyberine research. The orchestrator decomposes the question...","icon":"$undefined","url":"/docs/usage/run-recipes/research-mode--cyberine-research-plan---search---verify---synthesize","$ref":"usage/run-recipes/research-mode--cyberine-research-plan---search---verify---synthesize.md"},{"$id":"usage/run-recipes/prior-art-research--cyberine-research---prior-art-ks0sh.md","type":"page","name":"Prior-art research — cyberine research --prior-art (ks0sh)","description":"Competitor and alternatives research that ends in a feature parity matrix and a recommendations ledger, where every recommendation leaves the report as a task id or a dated rejection reason.","icon":"$undefined","url":"/docs/usage/run-recipes/prior-art-research--cyberine-research---prior-art-ks0sh","$ref":"usage/run-recipes/prior-art-research--cyberine-research---prior-art-ks0sh.md"},{"$id":"usage/run-recipes/website-rebuild-target-nextjs--fumadocs--site.md","type":"page","name":"Website rebuild target (Next.js + Fumadocs) — site/","description":"The rebuild (T666) lives in site/ as a separate buildable target, and since the T670 cutover (2026-09-25) it is the only docs site.","icon":"$undefined","url":"/docs/usage/run-recipes/website-rebuild-target-nextjs--fumadocs--site","$ref":"usage/run-recipes/website-rebuild-target-nextjs--fumadocs--site.md"},{"$id":"usage/run-recipes/documentation-site-vitepress-retired-t670.md","type":"page","name":"Documentation site: VitePress retired (T670)","description":"The VitePress implementation (docs/.vitepress, npm run docs:legacy:) was deleted on 2026-09-25 after the new site passed its full QA gate: the static site:qa pass (0 HIGH, 0 dead links across 59...","icon":"$undefined","url":"/docs/usage/run-recipes/documentation-site-vitepress-retired-t670","$ref":"usage/run-recipes/documentation-site-vitepress-retired-t670.md"},{"$id":"usage/run-recipes/fleet-policy-file--cyberine-fleet-t422-t425-t426.md","type":"page","name":"Fleet policy file — cyberine fleet (T422, T425, T426)","description":"fleet.json is an optional provider policy file that declares which workers are active and which model handles each task kind. It externalizes the routing that is otherwise hardcoded across worker...","icon":"$undefined","url":"/docs/usage/run-recipes/fleet-policy-file--cyberine-fleet-t422-t425-t426","$ref":"usage/run-recipes/fleet-policy-file--cyberine-fleet-t422-t425-t426.md"},{"$id":"usage/run-recipes/unified-health-check--cyberine-doctor-t540.md","type":"page","name":"Unified health check — cyberine doctor (T540)","description":"One report covering the whole install, printed as pass/warn/fail lines with a fix hint on every warn/fail:","icon":"$undefined","url":"/docs/usage/run-recipes/unified-health-check--cyberine-doctor-t540","$ref":"usage/run-recipes/unified-health-check--cyberine-doctor-t540.md"},{"$id":"usage/run-recipes/adapter-liveness--is-it-funded-not-just-keyed---liveness-kwrp9.md","type":"page","name":"Adapter liveness — is it FUNDED, not just keyed (--liveness, kwrp9)","description":"cyberine doctor answers \"is a key present\". It never answered \"is that key's account funded\", and neither did the fleet TUI's overview, which listed an adapter whenever its key existed and printed...","icon":"$undefined","url":"/docs/usage/run-recipes/adapter-liveness--is-it-funded-not-just-keyed---liveness-kwrp9","$ref":"usage/run-recipes/adapter-liveness--is-it-funded-not-just-keyed---liveness-kwrp9.md"},{"$id":"usage/run-recipes/selftest-sweep-with-a-real-skip-channel--npm-run-sweep-t707.md","type":"page","name":"Selftest sweep with a real SKIP channel — npm run sweep (T707)","description":"This project has no test framework; the selftest: smokes ARE the test suite, and the sweep is how you run all of them, on the operator's word, typically before a promotion. It replaced ~15 lines...","icon":"$undefined","url":"/docs/usage/run-recipes/selftest-sweep-with-a-real-skip-channel--npm-run-sweep-t707","$ref":"usage/run-recipes/selftest-sweep-with-a-real-skip-channel--npm-run-sweep-t707.md"},{"$id":"usage/run-recipes/security-gate-drift-check--npm-run-selftestgate-list-drift-t707-retargeted-2026-09-07.md","type":"page","name":"Security-gate drift check — npm run selftest:gate-list-drift (T707, retargeted 2026-09-07)","description":"The security/data-integrity smoke list froze on 2026-08-22 in two hand-maintained copies (the CI gate loop and the git-workflow rule's precondition 3), and ten security smokes shipped afterwards...","icon":"$undefined","url":"/docs/usage/run-recipes/security-gate-drift-check--npm-run-selftestgate-list-drift-t707-retargeted-2026-09-07","$ref":"usage/run-recipes/security-gate-drift-check--npm-run-selftestgate-list-drift-t707-retargeted-2026-09-07.md"},{"$id":"usage/run-recipes/promotion-check--npm-run-promotecheck-2026-10-02.md","type":"page","name":"Promotion check — npm run promote:check (2026-10-02)","description":"Operator decision 2026-10-02: no customers yet and still in the dev phase, so there is no automatic CI. .github/workflows/ci.yml and .githooks/pre-push were deleted, the homelab self-hosted runner...","icon":"$undefined","url":"/docs/usage/run-recipes/promotion-check--npm-run-promotecheck-2026-10-02","$ref":"usage/run-recipes/promotion-check--npm-run-promotecheck-2026-10-02.md"},{"$id":"usage/run-recipes/sweep-stamp--localselftest-stamp-2026-09-07-push-gate-removed-2026-09-30-hook-removed-2026-10-02.md","type":"page","name":"Sweep stamp — .local/selftest-stamp/ (2026-09-07; push gate removed 2026-09-30, hook removed 2026-10-02)","description":"A full npm run sweep (no --only / --security) records its verdict in .local/selftest-stamp/<host>/<HEAD sha>.json. From 2026-09-07 a tracked pre-push hook refused a push to main without an ok:...","icon":"$undefined","url":"/docs/usage/run-recipes/sweep-stamp--localselftest-stamp-2026-09-07-push-gate-removed-2026-09-30-hook-removed-2026-10-02","$ref":"usage/run-recipes/sweep-stamp--localselftest-stamp-2026-09-07-push-gate-removed-2026-09-30-hook-removed-2026-10-02.md"},{"$id":"usage/run-recipes/sweep-flake-ledger--npm-run-sweepflakes-kdpxk-2026-09-28.md","type":"page","name":"Sweep flake ledger — npm run sweep:flakes (kdpxk, 2026-09-28)","description":"Every sweep, full or partial, appends one line to .local/selftest-stamp/<host>/ledger.jsonl: at, sha, host, scope (full / security / only, with selected for a partial run), the pass count,...","icon":"$undefined","url":"/docs/usage/run-recipes/sweep-flake-ledger--npm-run-sweepflakes-kdpxk-2026-09-28","$ref":"usage/run-recipes/sweep-flake-ledger--npm-run-sweepflakes-kdpxk-2026-09-28.md"},{"$id":"usage/run-recipes/dependency-audit-gate--npm-run-selftestdep-audit-t673-multi-tree-since-t707.md","type":"page","name":"Dependency audit gate — npm run selftest:dep-audit (T673, multi-tree since T707)","description":"Recurring guard against silently-accruing vulnerable dependencies (the gap that let fast-uri / ip-address / nanoid / postcss sit at HIGH for weeks before 2026-08-20). It discovers every...","icon":"$undefined","url":"/docs/usage/run-recipes/dependency-audit-gate--npm-run-selftestdep-audit-t673-multi-tree-since-t707","$ref":"usage/run-recipes/dependency-audit-gate--npm-run-selftestdep-audit-t673-multi-tree-since-t707.md"},{"$id":"usage/run-recipes/credit-ledger--track-provider-balances-gate-low-credit-workers.md","type":"page","name":"Credit ledger — track provider balances, gate low-credit workers","description":"Cyberine can track an approximate estimated remaining balance per provider and skip or down-rank workers whose provider is running low. Credits do not need to match the provider dashboard exactly;...","icon":"$undefined","url":"/docs/usage/run-recipes/credit-ledger--track-provider-balances-gate-low-credit-workers","$ref":"usage/run-recipes/credit-ledger--track-provider-balances-gate-low-credit-workers.md"},{"$id":"usage/run-recipes/tui-stdout-shim--synchronized-output-and-how-to-smoke-it.md","type":"page","name":"TUI stdout shim — synchronized output, and how to smoke it","description":"Ink repaints the whole frame every tick. makeSyncStdout() (src/tui/syncOutput.ts) proxies the render stream and wraps each write in the DEC 2026 begin/end synchronized-update markers, so a...","icon":"$undefined","url":"/docs/usage/run-recipes/tui-stdout-shim--synchronized-output-and-how-to-smoke-it","$ref":"usage/run-recipes/tui-stdout-shim--synchronized-output-and-how-to-smoke-it.md"},{"$id":"usage/run-recipes/task-footprint-estimator--history-driven-budget-planning-kx9rt-calibrated-in-k5m89.md","type":"page","name":"Task footprint estimator — history-driven budget planning (kx9rt, calibrated in k5m89)","description":"planBudget() decides whether a step can spawn all its workers, must chunk into a handoff chain, should degrade, or has to abort. Its cost side comes from the credit ledger above; its SIZE side...","icon":"$undefined","url":"/docs/usage/run-recipes/task-footprint-estimator--history-driven-budget-planning-kx9rt-calibrated-in-k5m89","$ref":"usage/run-recipes/task-footprint-estimator--history-driven-budget-planning-kx9rt-calibrated-in-k5m89.md"},{"$id":"usage/run-recipes/usage-records-for-external-meters--billed-dollars-per-model-tokens-aux-spend-k0nnx-kq39p-kcajh.md","type":"page","name":"Usage records for external meters — billed dollars, per-model tokens, aux spend (k0nnx, kq39p, kcajh)","description":"Every dollar figure Cyberine keeps now travels with the model and the token counts behind it, so a meter outside Cyberine (tokenburn, a spreadsheet) can price it again or check it without...","icon":"$undefined","url":"/docs/usage/run-recipes/usage-records-for-external-meters--billed-dollars-per-model-tokens-aux-spend-k0nnx-kq39p-kcajh","$ref":"usage/run-recipes/usage-records-for-external-meters--billed-dollars-per-model-tokens-aux-spend-k0nnx-kq39p-kcajh.md"},{"$id":"usage/run-recipes/hard-spend-cap--opt-in-race-free-ceiling-across-all-providers.md","type":"page","name":"Hard spend cap — opt-in, race-free ceiling across ALL providers","description":"CYBERINECREDITMINUSD above is a SOFT, per-provider gate: it reads the folded balance and down-ranks/skips a worker, but two concurrent dispatches can each read \"balance ok\" moments apart and,...","icon":"$undefined","url":"/docs/usage/run-recipes/hard-spend-cap--opt-in-race-free-ceiling-across-all-providers","$ref":"usage/run-recipes/hard-spend-cap--opt-in-race-free-ceiling-across-all-providers.md"},{"$id":"usage/run-recipes/quota-cooldown--automatic-circuit-breaker-after-402429.md","type":"page","name":"Quota cooldown — automatic circuit-breaker after 402/429","description":"When a worker fails with a quota signal (HTTP 402 payment required / insufficient credits, or HTTP 429 rate-limit / too-many-requests), Cyberine records a temporary cooldown for that worker ID in...","icon":"$undefined","url":"/docs/usage/run-recipes/quota-cooldown--automatic-circuit-breaker-after-402429","$ref":"usage/run-recipes/quota-cooldown--automatic-circuit-breaker-after-402429.md"},{"$id":"usage/run-recipes/blocker-visibility--waitingfor-on-stalled-workers.md","type":"page","name":"Blocker visibility — waitingFor on stalled workers","description":"When a worker cannot make progress, a waitingFor field explains why. It appears in:","icon":"$undefined","url":"/docs/usage/run-recipes/blocker-visibility--waitingfor-on-stalled-workers","$ref":"usage/run-recipes/blocker-visibility--waitingfor-on-stalled-workers.md"},{"$id":"usage/run-recipes/provider-circuit-breaker-and-bounded-retry--one-dead-provider-degrades-the-fleet-not-the-run.md","type":"page","name":"Provider circuit breaker and bounded retry — one dead provider degrades the fleet, not the run","description":"Distinct from the quota cooldown above, which reacts only to a 402/429 quota signal: the breaker counts consecutive terminal failures of any shape at the Orchestrator.dispatch() seam, so it also...","icon":"$undefined","url":"/docs/usage/run-recipes/provider-circuit-breaker-and-bounded-retry--one-dead-provider-degrades-the-fleet-not-the-run","$ref":"usage/run-recipes/provider-circuit-breaker-and-bounded-retry--one-dead-provider-degrades-the-fleet-not-the-run.md"},{"$id":"usage/run-recipes/fallback-model-chain--per-worker-model-degradation.md","type":"page","name":"Fallback model chain — per-worker model degradation","description":"When a LoopAgentAdapter worker's primary model is overloaded or unavailable (HTTP 429 / 402 / 5xx / unexpected overload 400), Cyberine retries the same request with the next model in a configured...","icon":"$undefined","url":"/docs/usage/run-recipes/fallback-model-chain--per-worker-model-degradation","$ref":"usage/run-recipes/fallback-model-chain--per-worker-model-degradation.md"},{"$id":"usage/run-recipes/cli-flag-compat-diagnosis--structured-output-knobs-t656.md","type":"page","name":"CLI flag-compat diagnosis + structured-output knobs (T656)","description":"The claude and codex build workers pass version-gated CLI flags (e.g. codex's --json, codex's --dangerously-bypass-hook-trust when CYBERINECODEXBYPASSHOOKTRUST=1 opts in, claude's...","icon":"$undefined","url":"/docs/usage/run-recipes/cli-flag-compat-diagnosis--structured-output-knobs-t656","$ref":"usage/run-recipes/cli-flag-compat-diagnosis--structured-output-knobs-t656.md"},{"$id":"usage/run-recipes/claude-cold-spawn-session-persistence-suppression-t627.md","type":"page","name":"Claude cold-spawn session-persistence suppression (T627)","description":"Every cold-spawn claude build-worker dispatch left a resumable session transcript on disk under the operator's own ~/.claude session store, even though Cyberine never resumes a cold-spawn worker —...","icon":"$undefined","url":"/docs/usage/run-recipes/claude-cold-spawn-session-persistence-suppression-t627","$ref":"usage/run-recipes/claude-cold-spawn-session-persistence-suppression-t627.md"},{"$id":"usage/run-recipes/claude-prompt-answering----permission-prompts-none-k7hha.md","type":"page","name":"Claude prompt answering — --permission-prompts none (k7hha)","description":"A claude build worker is spawned with stdin CLOSED and with neither an SDK host nor --permission-prompt-tool. Claude Code's --permission-prompts flag (v2.1.259) defaults to host, which routes any...","icon":"$undefined","url":"/docs/usage/run-recipes/claude-prompt-answering----permission-prompts-none-k7hha","$ref":"usage/run-recipes/claude-prompt-answering----permission-prompts-none-k7hha.md"},{"$id":"usage/run-recipes/a-non-success-result-subtype-is-a-failed-run-kq5rd.md","type":"page","name":"A non-success result subtype is a failed run (kq5rd)","description":"parseClaudeStreamJson used to take the last type:'result' event and read its result field without looking at subtype. An error subtype carries no result (Claude Code puts errors[] there instead),...","icon":"$undefined","url":"/docs/usage/run-recipes/a-non-success-result-subtype-is-a-failed-run-kq5rd","$ref":"usage/run-recipes/a-non-success-result-subtype-is-a-failed-run-kq5rd.md"},{"$id":"usage/run-recipes/where-a-workers-claude_code_-settings-came-from--toolingsnapshotworkerenv-kg4pt.md","type":"page","name":"Where a worker's CLAUDE_CODE_* settings came from — toolingSnapshot.workerEnv (kg4pt)","description":"A spawned claude worker inherits the orchestrator's environment verbatim apart from provider credentials (buildScopedChildEnv). A Claude Code session exports its settings.json env block, so a...","icon":"$undefined","url":"/docs/usage/run-recipes/where-a-workers-claude_code_-settings-came-from--toolingsnapshotworkerenv-kg4pt","$ref":"usage/run-recipes/where-a-workers-claude_code_-settings-came-from--toolingsnapshotworkerenv-kg4pt.md"},{"$id":"usage/run-recipes/validate-a-harness-overlay--cyberine-harness-check-kdgkp.md","type":"page","name":"Validate a harness overlay — cyberine harness check (kdgkp)","description":"A harness overlay is a directory of proposed edits to what Cyberine injects into workers: rules/<name>.md (replaces or adds a project rule file in .cyberine/rules/), skills/<id>/SKILL.md and files...","icon":"$undefined","url":"/docs/usage/run-recipes/validate-a-harness-overlay--cyberine-harness-check-kdgkp","$ref":"usage/run-recipes/validate-a-harness-overlay--cyberine-harness-check-kdgkp.md"},{"$id":"usage/run-recipes/one-harness-repair-after-a-verify-failure--cyberine_harness_repair-k5dc7.md","type":"page","name":"One harness repair after a verify failure — CYBERINE_HARNESS_REPAIR (k5dc7)","description":"Off by default (operator decision D3-A). With CYBERINEHARNESSREPAIR=1, a code task that cyberine autopilot runs on its fleet engine with sync dispatch (the default; --dispatch sync or --verify)...","icon":"$undefined","url":"/docs/usage/run-recipes/one-harness-repair-after-a-verify-failure--cyberine_harness_repair-k5dc7","$ref":"usage/run-recipes/one-harness-repair-after-a-verify-failure--cyberine_harness_repair-k5dc7.md"},{"$id":"usage/run-recipes/contrastive-reflection-into-a-lesson-library--cyberine-lessons-k79pw.md","type":"page","name":"Contrastive reflection into a lesson library — cyberine lessons (k79pw)","description":"Reflect on one completed step only when the record contains a passed worker and a failed worker with the same brief. A judge selecting one of several successful workers is not a failure signal....","icon":"$undefined","url":"/docs/usage/run-recipes/contrastive-reflection-into-a-lesson-library--cyberine-lessons-k79pw","$ref":"usage/run-recipes/contrastive-reflection-into-a-lesson-library--cyberine-lessons-k79pw.md"},{"$id":"usage/run-recipes/which-harness-a-worker-ran-with--toolingsnapshotharness-k7xxg.md","type":"page","name":"Which harness a worker ran with — toolingSnapshot.harness (k7xxg)","description":"Every engine run records, in run.json, a hash of each harness input Cyberine injected, so two runs can be compared on the harness they used without storing the text itself (instruction files can...","icon":"$undefined","url":"/docs/usage/run-recipes/which-harness-a-worker-ran-with--toolingsnapshotharness-k7xxg","$ref":"usage/run-recipes/which-harness-a-worker-ran-with--toolingsnapshotharness-k7xxg.md"},{"$id":"usage/run-recipes/which-claude-account-a-worker-bills--oauthtoken-k8rhz.md","type":"page","name":"Which Claude account a worker bills — oauthToken (k8rhz)","description":"CLAUDECODEOAUTHTOKEN is a long-lived subscription credential that decides WHICH SEAT a claude worker bills. It used to reach every spawned worker by plain inheritance: T603's least-privilege env","icon":"$undefined","url":"/docs/usage/run-recipes/which-claude-account-a-worker-bills--oauthtoken-k8rhz","$ref":"usage/run-recipes/which-claude-account-a-worker-bills--oauthtoken-k8rhz.md"},{"$id":"usage/run-recipes/which-secrets-a-cli-worker-sees--generic-secret-strip-and-passthrough-kznbp.md","type":"page","name":"Which secrets a CLI worker sees — generic secret strip and passthrough (kznbp)","description":"T603 and k8rhz scoped provider credentials only, so every other secret in the launching shell reached an auto-approve CLI worker: GITHUBTOKEN, NPMTOKEN, AWSSECRETACCESSKEY, DATABASEURL,...","icon":"$undefined","url":"/docs/usage/run-recipes/which-secrets-a-cli-worker-sees--generic-secret-strip-and-passthrough-kznbp","$ref":"usage/run-recipes/which-secrets-a-cli-worker-sees--generic-secret-strip-and-passthrough-kznbp.md"},{"$id":"usage/run-recipes/claude-session-names--cy-action-run8-session8-seat.md","type":"page","name":"Claude session names — cy-<action>-<run8>-<session8>[-<seat>]","description":"Every claude session Cyberine spawns registers under a name that starts with cy-, so claude agents, ListAgents and ~/.claude/sessions/<pid>.json tell a Cyberine child apart from the operator's own...","icon":"$undefined","url":"/docs/usage/run-recipes/claude-session-names--cy-action-run8-session8-seat","$ref":"usage/run-recipes/claude-session-names--cy-action-run8-session8-seat.md"},{"$id":"usage/run-recipes/background-fleet-worker--claude---bg-k56eq.md","type":"page","name":"Background fleet worker — claude --bg (k56eq)","description":"The default claude build worker is an opaque claude -p child: nothing can open it, steer it, or outlive the orchestrator that spawned it. CYBERINECLAUDEWORKERMODE=bg swaps that ONE worker for a...","icon":"$undefined","url":"/docs/usage/run-recipes/background-fleet-worker--claude---bg-k56eq","$ref":"usage/run-recipes/background-fleet-worker--claude---bg-k56eq.md"},{"$id":"usage/run-recipes/idle-stream-watchdog--abort-a-stalled-stream-fleet-worker--chat-t562-t570.md","type":"page","name":"Idle-stream watchdog — abort a stalled stream, fleet worker + chat (T562, T570)","description":"Claude Code v2.1.196 made a streaming idle watchdog default-on: it aborts and retries a response stream that produces no events for 5 minutes. LoopAgentAdapter (fleet workers) AND the interactive...","icon":"$undefined","url":"/docs/usage/run-recipes/idle-stream-watchdog--abort-a-stalled-stream-fleet-worker--chat-t562-t570","$ref":"usage/run-recipes/idle-stream-watchdog--abort-a-stalled-stream-fleet-worker--chat-t562-t570.md"},{"$id":"usage/run-recipes/worker-internal-subagents--delegate_subtask-t313.md","type":"page","name":"Worker-internal subagents — delegate_subtask (T313)","description":"Implementer-class loop workers (loop--build) automatically get a delegatesubtask tool that lets the worker spawn a scoped subagent — a focused nested LoopAgentAdapter with its own system prompt, a...","icon":"$undefined","url":"/docs/usage/run-recipes/worker-internal-subagents--delegate_subtask-t313","$ref":"usage/run-recipes/worker-internal-subagents--delegate_subtask-t313.md"},{"$id":"usage/run-recipes/tool-call-pipeline--batch-multi-tool-sequences-in-one-model-turn-t319.md","type":"page","name":"Tool-call pipeline — batch multi-tool sequences in one model turn (T319)","description":"Implementer-class loop workers (loop--build) automatically get a runpipeline tool that lets a worker express a predictable multi-step tool sequence as a single tool call, collapsing N sequential...","icon":"$undefined","url":"/docs/usage/run-recipes/tool-call-pipeline--batch-multi-tool-sequences-in-one-model-turn-t319","$ref":"usage/run-recipes/tool-call-pipeline--batch-multi-tool-sequences-in-one-model-turn-t319.md"},{"$id":"usage/run-recipes/result-projection--configurable-worker-output-fidelity-t328.md","type":"page","name":"Result projection — configurable worker output fidelity (T328)","description":"By default every worker returns a full Result: the complete output text, usage stats, and any raw provider payload. When a judge or downstream step only needs a summary (e.g. an LLM judge that...","icon":"$undefined","url":"/docs/usage/run-recipes/result-projection--configurable-worker-output-fidelity-t328","$ref":"usage/run-recipes/result-projection--configurable-worker-output-fidelity-t328.md"},{"$id":"usage/run-recipes/stateful-session-reuse--warm-cli-agent-bindings--session-tree-ingestion-t333.md","type":"page","name":"Stateful session reuse — warm CLI-agent bindings + session tree ingestion (T333)","description":"By default every CLI worker cold-spawns a brand-new session for every task. With warm bindings enabled the adapter records the session ID returned by the first successful dispatch, then passes it...","icon":"$undefined","url":"/docs/usage/run-recipes/stateful-session-reuse--warm-cli-agent-bindings--session-tree-ingestion-t333","$ref":"usage/run-recipes/stateful-session-reuse--warm-cli-agent-bindings--session-tree-ingestion-t333.md"},{"$id":"usage/run-recipes/pluggable-execution-backends--run-workers-in-docker-or-on-a-remote-host-t332.md","type":"page","name":"Pluggable execution backends — run workers in Docker or on a remote host (T332)","description":"By default every isolate-step worker provisions a local git worktree and executes its CLI or loop work on the local host. The execution backend abstraction decouples WHERE the worker runs from...","icon":"$undefined","url":"/docs/usage/run-recipes/pluggable-execution-backends--run-workers-in-docker-or-on-a-remote-host-t332","$ref":"usage/run-recipes/pluggable-execution-backends--run-workers-in-docker-or-on-a-remote-host-t332.md"}],"$id":"usage/run-recipes","$ref":{"folder":"usage/run-recipes","meta":"usage/run-recipes/meta.json"},"icon":"$undefined"},{"$id":"usage/extend-recipes-the-vendor-independence-payoff.md","type":"page","name":"Extend recipes (the vendor-independence payoff)","description":"Three independent ways, none touch the core:","icon":"$undefined","url":"/docs/usage/extend-recipes-the-vendor-independence-payoff","$ref":"usage/extend-recipes-the-vendor-independence-payoff.md"},{"$id":"usage/durable-teammate.md","type":"page","name":"Durable teammate","description":"Enables two complementary approaches so a teammate can receive and act on messages even after it has gone idle or exited. Both share the same core handler (src/store/durableTeammate.ts). Default...","icon":"$undefined","url":"/docs/usage/durable-teammate","$ref":"usage/durable-teammate.md"},{"$id":"usage/cross-run-comms-over-http-hub-and-spoke-message-bus.md","type":"page","name":"Cross-run comms over HTTP (hub-and-spoke message bus)","description":"The default file bus backend is filesystem IPC: it reaches every process on ONE machine and nothing beyond it. CYBERINEBUSBACKEND=http swaps in a backend that carries the same publish / subscribe...","icon":"$undefined","url":"/docs/usage/cross-run-comms-over-http-hub-and-spoke-message-bus","$ref":"usage/cross-run-comms-over-http-hub-and-spoke-message-bus.md"},{"$id":"usage/background-shell-jobs-inside-a-worker-run_bash-background--bash_jobs.md","type":"page","name":"Background shell jobs inside a worker (run_bash background + bash_jobs)","description":"A loop worker can start a long-running shell command without blocking its tool round: runbash with background: true starts the command immediately and returns a job id; bashjobs with action=\"list\"...","icon":"$undefined","url":"/docs/usage/background-shell-jobs-inside-a-worker-run_bash-background--bash_jobs","$ref":"usage/background-shell-jobs-inside-a-worker-run_bash-background--bash_jobs.md"},{"$id":"usage/what-cyberine-writes-on-disk.md","type":"page","name":"What Cyberine writes on disk","description":"A reference of every path the product code creates, built from the writers in src/ (2026-10-18). \"Secret\" marks a file that can hold credentials or private text: keep it out of git, backups you...","icon":"$undefined","url":"/docs/usage/what-cyberine-writes-on-disk","$ref":"usage/what-cyberine-writes-on-disk.md"},{"$id":"usage/environment-variables.md","type":"page","name":"Environment variables","description":"Environment variables: a Cyberine usage recipe.","icon":"$undefined","url":"/docs/usage/environment-variables","$ref":"usage/environment-variables.md"},{"$id":"usage/generated-reference-do-not-edit-by-hand.md","type":"page","name":"Generated reference (do not edit by hand)","description":"Regenerate with npm run docs:catalog. npm run docs:catalog:check fails (non-zero exit, writes nothing) when source has drifted from these tables -- e.g. a new tool or env var was added and this...","icon":"$undefined","url":"/docs/usage/generated-reference-do-not-edit-by-hand","$ref":"usage/generated-reference-do-not-edit-by-hand.md"}],"$id":"usage","$ref":{"folder":"usage","meta":"usage/meta.json"},"icon":"$undefined"},{"$id":"tasks-manifest.md","type":"page","name":"Task manifest format","description":"This is the normative specification of the task-manifest format used by the cyberine orchestrator (src/tasks/manifestReader.ts, src/tasks/statusWriter.ts, src/tasks/runnable.ts on the read side;...","icon":"$undefined","url":"/docs/tasks-manifest","$ref":"tasks-manifest.md"},{"type":"folder","name":"Providers","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"providers/index.md","type":"page","name":"Provider API Reference","description":"A trace-it-fast reference for every AI provider / coding-CLI that Cyberine integrates directly. Each provider has its own page with the official API/CLI surface (endpoints, auth, models, gotchas,...","icon":"$undefined","url":"/docs/providers","$ref":"providers/index.md"},{"$id":"providers/changelog-watch.md","type":"page","name":"Upstream Changelog Watch List","description":"Every upstream Cyberine follows, in one table, so \"anything new upstream?\" resolves to the right changelog in one hop. URLs are pulled from each provider's page in this folder (its \"Official docs\"...","icon":"$undefined","url":"/docs/providers/changelog-watch","$ref":"providers/changelog-watch.md"},{"$id":"providers/model-strength.md","type":"page","name":"Model strength reference (2026-09-17)","description":"One sourced answer to \"which enabled model is strongest\", used by the router today and meant for planning, task splitting and judging. It ranks every model id the fleet can run (every...","icon":"$undefined","url":"/docs/providers/model-strength","$ref":"providers/model-strength.md"},{"$id":"providers/deepseek.md","type":"page","name":"DeepSeek API (direct)","description":"All paths are relative to https://api.deepseek.com. Every one below was called live on 2026-08-25 (T679) and returned the status shown.","icon":"$undefined","url":"/docs/providers/deepseek","$ref":"providers/deepseek.md"},{"$id":"providers/openrouter.md","type":"page","name":"OpenRouter API","description":"No per-provider auth — OpenRouter wraps every provider under one key.","icon":"$undefined","url":"/docs/providers/openrouter","$ref":"providers/openrouter.md"},{"$id":"providers/anthropic.md","type":"page","name":"Anthropic Claude API (Messages)","description":"Fable 5 $10/$50 · Opus 5.5 $4/$20 · Opus 5 + 4.x $5/$25 (Opus 5 / Opus 4.8 fast mode $10/$50) · Sonnet 5.5 and Sonnet 5 $2/$10 (permanent since 2026-08; the 2026-09-01 rise to $3/$15 was...","icon":"$undefined","url":"/docs/providers/anthropic","$ref":"providers/anthropic.md"},{"$id":"providers/openai.md","type":"page","name":"OpenAI API (Chat Completions + Responses)","description":"The openai Node SDK constructor accepts a custom baseURL — this is how Cyberine points the same client at LiteLLM/OpenRouter/DeepSeek without code changes.","icon":"$undefined","url":"/docs/providers/openai","$ref":"providers/openai.md"},{"$id":"providers/litellm.md","type":"page","name":"LiteLLM Proxy / Gateway","description":"These two defaults are placeholders that signal \"unconfigured\" — override them with real upstream ids. Underlying provider docs: openai.md, anthropic.md.","icon":"$undefined","url":"/docs/providers/litellm","$ref":"providers/litellm.md"},{"$id":"providers/moonshot-kimi.md","type":"page","name":"Moonshot AI / Kimi API","description":"A vendor-independent multi-provider AI agent orchestrator: dispatch one task to many agents, judge the results, ship the winner.","icon":"$undefined","url":"/docs/providers/moonshot-kimi","$ref":"providers/moonshot-kimi.md"},{"$id":"providers/minimax.md","type":"page","name":"MiniMax (direct API)","description":"Fact-checked 2026-06-10 against https://platform.minimax.io/docs/api-reference/text-openai-api.","icon":"$undefined","url":"/docs/providers/minimax","$ref":"providers/minimax.md"},{"$id":"providers/command-code.md","type":"page","name":"Command Code (CLI + Provider API)","description":"Reference for Command Code (commandcode.ai): the cmd coding CLI and its OpenAI-compatible Provider API. Read this before any Command Code adapter, headroom or fleet work.","icon":"$undefined","url":"/docs/providers/command-code","$ref":"providers/command-code.md"},{"$id":"providers/ollama.md","type":"page","name":"Ollama (local models)","description":"Drop-in for the existing LoopAgentAdapter (openai SDK): swap baseURL, static apiKey. Empirically verified on v0.30.6:","icon":"$undefined","url":"/docs/providers/ollama","$ref":"providers/ollama.md"},{"$id":"providers/ollama-cloud.md","type":"page","name":"Ollama Cloud (hosted inference)","description":"Fact-checked 2026-06-27 against https://docs.ollama.com/cloud and https://ollama.com/blog/cloud-models; cross-referenced with https://fabiorehm.com/blog/2026/04/12/pi-ollama-cloud-api/ (hands-on...","icon":"$undefined","url":"/docs/providers/ollama-cloud","$ref":"providers/ollama-cloud.md"},{"$id":"providers/direct-api.md","type":"page","name":"Direct-API Providers (OpenAI-compatible key-loop)","description":"This page covers the OpenAI-compatible direct-key providers registered in Cyberine as loop--build build-fleet workers. Each uses LoopAgentAdapter and requires only an API key — no CLI install, no...","icon":"$undefined","url":"/docs/providers/direct-api","$ref":"providers/direct-api.md"},{"$id":"providers/cloud-native.md","type":"page","name":"Cloud-Native Providers (Bedrock, Vertex AI)","description":"These two workers use cloud-platform auth rather than a plain API key. None of them routes through the LiteLLM gateway; each has a dedicated adapter in src/adapters/.","icon":"$undefined","url":"/docs/providers/cloud-native","$ref":"providers/cloud-native.md"},{"$id":"providers/cli-agents.md","type":"page","name":"CLI Coding Agents — adapter integration reference","description":"Pre-crawled, fact-checked reference for adding headless CLI coding agents as Cyberine workers (CliAgentAdapter spawns the official binary non-interactively, parses stdout into a Result)....","icon":"$undefined","url":"/docs/providers/cli-agents","$ref":"providers/cli-agents.md"},{"$id":"providers/claude-code-cli.md","type":"page","name":"Claude Code CLI (headless)","description":"Default model: claude-sonnet-5-5 (CYBERINECLAUDEMODEL; was claude-sonnet-5 until 2026-10-01, kvhxs). Auth dir allow-listed by the sandbox: ~/.claude.","icon":"$undefined","url":"/docs/providers/claude-code-cli","$ref":"providers/claude-code-cli.md"},{"$id":"providers/codex-cli.md","type":"page","name":"OpenAI Codex CLI (headless)","description":"Model: CYBERINECODEXMODEL if set, else Codex's configured default (no -m added). Deliberately unpinned (kd7ax, 2026-09-09): Codex 0.153.4 made GPT-6-Astra the bundled default, and Cyberine keeps...","icon":"$undefined","url":"/docs/providers/codex-cli","$ref":"providers/codex-cli.md"},{"$id":"providers/gemini-cli.md","type":"page","name":"Google Gemini CLI (headless)","description":"Default model: gemini-2.5-flash (CYBERINEGEMINIMODEL / GEMINIMODEL). Auth dir allow-listed by the sandbox: ~/.gemini.","icon":"$undefined","url":"/docs/providers/gemini-cli","$ref":"providers/gemini-cli.md"},{"$id":"providers/opencode-cli.md","type":"page","name":"OpenCode CLI (anomalyco/opencode, headless)","description":"Verified 2026-08-25 (hl-t693). opencode has two independent billing surfaces, and only ONE of them died. Getting this wrong in either direction is expensive, so read both halves:","icon":"$undefined","url":"/docs/providers/opencode-cli","$ref":"providers/opencode-cli.md"},{"$id":"providers/opencode2-v2-beta.md","type":"page","name":"OpenCode 2.0 beta (opencode2) — integrated as opencode2-deepseek, currently retired","description":"v2 is unreachable from every surface CHANGELOG-WATCH.md's OpenCode row polls, which is the whole reason a user report (\"opencode ra version 2 roi\") beat the sweep to it:","icon":"$undefined","url":"/docs/providers/opencode2-v2-beta","$ref":"providers/opencode2-v2-beta.md"},{"$id":"providers/muse-cli.md","type":"page","name":"Muse Code (Meta) — headless muse exec","description":"Every claim below was checked against the installed binary or a real dispatch. Where a measurement contradicted the expectation, the measurement is what is recorded and the expectation is named so...","icon":"$undefined","url":"/docs/providers/muse-cli","$ref":"providers/muse-cli.md"},{"$id":"providers/mcp-2.0-audit.md","type":"page","name":"MCP Spec RC (2026-07-28) + SDK 2.0.0-beta — Audit (T639)","description":"Audit-only. No code changed, no dependency bumped. Scope: src/mcp.ts (MCP server — exposes the orchestrator) and src/adapters/McpToolAdapter.ts (MCP client — consumes arbitrary remote MCP...","icon":"$undefined","url":"/docs/providers/mcp-2.0-audit","$ref":"providers/mcp-2.0-audit.md"}],"$id":"providers","$ref":{"folder":"providers","meta":"providers/meta.json"},"icon":"$undefined"},{"type":"folder","name":"Design notes","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"design/website-master.md","type":"page","name":"Website MASTER — \"Run record\" (marketing) + \"Day Assembly / Night Approach\" (docs)","description":"Read this before any website work. Locked 2026-10-01 by the operator. It supersedes the Plasma home + \"Six Instruments\" routes MASTER, archived as website-master-20260925-plasma.md; the rejected...","icon":"$undefined","url":"/docs/design/website-master","$ref":"design/website-master.md"},{"$id":"design/website-spec.md","type":"page","name":"Website spec — implementation contract for \"Parallel Cut\"","description":"Companion to website-master.md, which owns the IDENTITY (palette, type, signature suite, motion vocabulary). This file owns everything MASTER deliberately does not carry: information architecture,...","icon":"$undefined","url":"/docs/design/website-spec","$ref":"design/website-spec.md"},{"$id":"design/website-theme-decision.md","type":"page","name":"Website theme decision - Dracula, dark-only; rebuild epic shelved","description":"Decided 2026-08-27 by the user, in one sitting, after the three-way color contradiction below was surfaced. This note exists so the question is not re-opened from scratch: the site's palette is...","icon":"$undefined","url":"/docs/design/website-theme-decision","$ref":"design/website-theme-decision.md"},{"$id":"design/website-master-20260925-plasma.md","type":"page","name":"ARCHIVED — Plasma home + \"Six Instruments\" routes (2026-08-27 to 2026-10-01)","description":"Read this before any website work. It replaces website-master-20260827-parallel-cut.md, which was rejected at prototype the same day it was written — read its post-mortem before proposing...","icon":"$undefined","url":"/docs/design/website-master-20260925-plasma","$ref":"design/website-master-20260925-plasma.md"},{"$id":"design/website-master-20260827-parallel-cut.md","type":"page","name":"ARCHIVED — \"Parallel Cut\" (rejected at prototype, 2026-08-27)","description":"The single source of truth for the Cyberine website's visual identity. Locked 2026-08-27 by the user via /nf-design-system heavy, after an independent three-way brainstorm (codex gpt-5.6-sol at...","icon":"$undefined","url":"/docs/design/website-master-20260827-parallel-cut","$ref":"design/website-master-20260827-parallel-cut.md"},{"$id":"design/persona.md","type":"page","name":"Persona — Cyberine","description":"Cyberine ships a single built-in persona, Cyberine, that gives the chat planner a consistent voice and identity. The persona is optional: set CYBERINEPERSONA=none to disable it and run with bare...","icon":"$undefined","url":"/docs/design/persona","$ref":"design/persona.md"},{"$id":"design/autonomy-loop.md","type":"page","name":"Autonomy Loop — T455 Design Note","description":"Status: design (T455). Locks the architecture for the AUTONOMY epic (T456-T466). No product code changed. Implementation tasks must treat the contracts here as fixed. Source cross-checked:...","icon":"$undefined","url":"/docs/design/autonomy-loop","$ref":"design/autonomy-loop.md"},{"$id":"design/loop-sync-dispatch-verify.md","type":"page","name":"Loop Synchronous Dispatch + Verify Seam — Design Note","description":"Status: shipped — T459 (beam-search forks) and T463 (non-code verify gate) are wired into the loop's syncDispatchTask path, see syncDispatchTask, forkBeamSearchCandidates and verifyNonCodeResult...","icon":"$undefined","url":"/docs/design/loop-sync-dispatch-verify","$ref":"design/loop-sync-dispatch-verify.md"},{"$id":"design/code-context-3act.md","type":"page","name":"Code-Context 3-Act Architecture — T440 Design","description":"Status: design (T440). Locks the architecture for the CODE-CONTEXT epic (T441-T454). No product code changed. Implementation tasks must treat the contracts here as fixed. Source cross-checked:...","icon":"$undefined","url":"/docs/design/code-context-3act","$ref":"design/code-context-3act.md"},{"$id":"design/long-term-memory.md","type":"page","name":"Long-term memory for Cyberine — survey + design","description":"Agent memory is organized into five cognitive categories — episodic (experiences situated in time), semantic (abstract facts), procedural (skills, policies), working (in-context manipulation),...","icon":"$undefined","url":"/docs/design/long-term-memory","$ref":"design/long-term-memory.md"},{"$id":"design/knowledge-store.md","type":"page","name":"Knowledge and plan store for Cyberine","description":"Shipped 2026-11-10 (tasks k6tbw, k8380, kgt0g, kcnvb, k493v, kx5xv, k9nhg, ksrat). This page describes what the code does now, read from the source at the time of writing. Where the original plans...","icon":"$undefined","url":"/docs/design/knowledge-store","$ref":"design/knowledge-store.md"},{"$id":"design/lsp-diagnostic-feedback-loop.md","type":"page","name":"LSP Diagnostic Feedback Loop into Adapters — T436 Design Note","description":"Status: shipped (T436) — see src/core/config.ts (diagFeedback, diagFeedbackInterval, diagSeverity, diagMaxErrors) and src/adapters/LoopAgentAdapter.ts (runDiagnostics, per-edit and end-of-loop...","icon":"$undefined","url":"/docs/design/lsp-diagnostic-feedback-loop","$ref":"design/lsp-diagnostic-feedback-loop.md"},{"$id":"design/fleet-config-design.md","type":"page","name":"Fleet Config — Design Draft","description":"fleet.json declares policy (who is on, which model for which task). It never holds API keys or secrets. Keys stay exactly where they are today:","icon":"$undefined","url":"/docs/design/fleet-config-design","$ref":"design/fleet-config-design.md"},{"$id":"design/tui-config-interaction-standard.md","type":"page","name":"TUI Config Interaction Standard — T418 Tier-1 Design Doc","description":"Status: design artifact only — no code built. Tiers 2-3 (shared primitives + migration + settings screen) are TTY+decision-gated and not included here. Source cross-checked: src/tui/,...","icon":"$undefined","url":"/docs/design/tui-config-interaction-standard","$ref":"design/tui-config-interaction-standard.md"},{"$id":"design/cli-flag-audit.md","type":"page","name":"CLI Flag Audit (T428)","description":"Inventory of every top-level subcommand and sub-subcommand, with --help/-h status before and after T428.","icon":"$undefined","url":"/docs/design/cli-flag-audit","$ref":"design/cli-flag-audit.md"},{"$id":"design/cli-streaming-output.md","type":"page","name":"CLI Streaming Output — T435 Design Note","description":"Status: shipped — Path 1 landed (T435). See src/orchestrator/Orchestrator.ts:144,374-375. Source cross-checked: src/commands/fleet.ts, src/cli.ts, src/orchestrator/run.ts,...","icon":"$undefined","url":"/docs/design/cli-streaming-output","$ref":"design/cli-streaming-output.md"},{"$id":"design/hook-matcher-anchoring.md","type":"page","name":"Hook Matcher: Anchoring vs Exact Match (T563)","description":"Claude Code v2.1.195 changed hook matchers to require an EXACT match for hyphenated identifiers (e.g. code-reviewer, mcpbrave-search). Previously CC's matcher ran an unanchored substring test, so...","icon":"$undefined","url":"/docs/design/hook-matcher-anchoring","$ref":"design/hook-matcher-anchoring.md"},{"$id":"design/bus-presence-auth-scope.md","type":"page","name":"Bus/Presence/MCP Auth Scope (T584)","description":"apiTokenIsValid (src/store/apiTokens.ts) is wired into exactly one place: the REST authorized() check in src/web/security.ts, which is itself skipped entirely when the web server is bound to...","icon":"$undefined","url":"/docs/design/bus-presence-auth-scope","$ref":"design/bus-presence-auth-scope.md"},{"$id":"design/contextq-integration.md","type":"page","name":"ContextQ Integration — Cyberine long-term memory on the Agent Memory Layer","description":"Status: design spec (drives implementation). Grounded against ContextQ as it ships today (shared-context-server, agent layer migration 0043) and Cyberine memory waves 1-4.","icon":"$undefined","url":"/docs/design/contextq-integration","$ref":"design/contextq-integration.md"},{"$id":"design/plugin-marketplace-decision.md","type":"page","name":"Plugin Marketplace / Discovery Registry — Build vs Defer (T542)","description":"This note decides ONE thing: whether Cyberine should build a hosted marketplace / discovery registry for plugins — a remote index a user can search/browse, with install-from-registry as a third...","icon":"$undefined","url":"/docs/design/plugin-marketplace-decision","$ref":"design/plugin-marketplace-decision.md"}],"$id":"design","$ref":{"folder":"design","meta":"design/meta.json"},"icon":"$undefined"},{"type":"folder","name":"Evaluation","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"eval/worker-profile-coverage.md","type":"page","name":"Worker-profile eval coverage (T678)","description":"Generated by npx tsx scripts/t678-profile-coverage.ts --write. Re-run it after any fleet dispatch to refresh; do not hand-edit.","icon":"$undefined","url":"/docs/eval/worker-profile-coverage","$ref":"eval/worker-profile-coverage.md"},{"$id":"eval/code-context-3act-eval.md","type":"page","name":"3-Act Code-Context Eval — T453","description":"Status: complete. Offline measurement only. No live LLM calls made. Harness: src/eval/contextEval.ts Design doc: docs/design/code-context-3act.md","icon":"$undefined","url":"/docs/eval/code-context-3act-eval","$ref":"eval/code-context-3act-eval.md"}],"$id":"eval","$ref":{"folder":"eval","meta":"eval/meta.json"},"icon":"$undefined"}],"fallback":{"type":"root","$ref":{"folder":"","meta":"$undefined"},"$id":"fallback:root","name":"Docs","description":"$undefined","children":[{"$id":"fallback:privacy.md","type":"page","name":"Privacy and licence","description":"Source reviewed: 2026-11-08. This note covers the Cyberine application and plugin, not the retention policies of third-party services or a website hosting provider.","icon":"$undefined","url":"/docs/privacy","$ref":"privacy.md"},{"type":"folder","name":"Design","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"fallback:design/board-as-sole-coordination-channel.md","type":"page","name":"The board is the only coordination channel","description":"Operator ruling, 2026-09-21. Binding on every design that touches how two Cyberine sessions divide work.","icon":"$undefined","url":"/docs/design/board-as-sole-coordination-channel","$ref":"design/board-as-sole-coordination-channel.md"},{"$id":"fallback:design/claude-accounts.md","type":"page","name":"Shared Claude accounts: claude-accounts/1","description":"Status: accepted by the operator 2026-10-14 (relayed by cc-session-float28), with D2 changed to a single root variable. cc-session-float implements every NEW item as of its main 5c6b379....","icon":"$undefined","url":"/docs/design/claude-accounts","$ref":"design/claude-accounts.md"},{"$id":"fallback:design/fanout-topology.md","type":"page","name":"Fan-out topology audit (kkxs4)","description":"Evidence: Google Research, \"Towards a science of scaling agent systems\" (2026-01-28). Independent multi-agent systems amplified errors 17.2x and centralised ones 4.4x. On sequential tasks...","icon":"$undefined","url":"/docs/design/fanout-topology","$ref":"design/fanout-topology.md"},{"$id":"fallback:design/guard-fail-closed.md","type":"page","name":"Board guard: fail open, not onFailure: \"block\" (kjtjc)","description":"Rejected, 2026-10-10. The plugin's PreToolUse board guard stays fail-open: plugin/hooks/hooks.json does not set onFailure: \"block\", and plugin/hooks/board-hooks.mjs keeps mapping every CLI failure...","icon":"$undefined","url":"/docs/design/guard-fail-closed","$ref":"design/guard-fail-closed.md"},{"$id":"fallback:design/handoff.md","type":"page","name":"Shared session handoff: handoff/1","description":"Status: version 1, published 2026-10-17 by Cyberine (ai-agent-solution145) at the operator's word (D1-A), answering cc-session-float's request on its task k373d (operator option D: one spec,...","icon":"$undefined","url":"/docs/design/handoff","$ref":"design/handoff.md"},{"$id":"fallback:design/harness-contract.md","type":"page","name":"Harness contract: one core, thin shells","description":"Status: accepted with edits, re-verified against the source 2026-10-28 (task ktagh)","icon":"$undefined","url":"/docs/design/harness-contract","$ref":"design/harness-contract.md"},{"$id":"fallback:design/tui-direction-a.md","type":"page","name":"TUI direction A: layout contract","description":"Status: as-built contract, re-verified against the source 2026-10-28 (task kqk3m) and updated 2026-10-29 (kk04j, kh72a, ks08k). It states what the source does now, per view and per breakpoint;...","icon":"$undefined","url":"/docs/design/tui-direction-a","$ref":"design/tui-direction-a.md"},{"$id":"fallback:design/tui-hub-blocks.md","type":"page","name":"TUI hub: blocks, inspector and motion","description":"Status: shipped 2026-10-04 (kp3fe, k2k49, kf366, ky347, k90w4, kjge7; documented by k1kzw). This is the hub section of the TUI design for the direction the operator picked from the 2026-10-04...","icon":"$undefined","url":"/docs/design/tui-hub-blocks","$ref":"design/tui-hub-blocks.md"},{"$id":"fallback:design/tui-keymap-architecture.md","type":"page","name":"TUI keymap architecture (KEYMAP-V2)","description":"Status: design, 2026-09-02. Derived from a read of two shipped harnesses (Claude Code 2.1.252, Codex 0.149.1) plus this repo's own web dashboard, which already ships the pattern the TUI lacks.","icon":"$undefined","url":"/docs/design/tui-keymap-architecture","$ref":"design/tui-keymap-architecture.md"}],"$id":"fallback:design","$ref":{"folder":"design","meta":"$undefined"},"icon":"$undefined"},{"type":"folder","name":"Providers","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"fallback:providers/cursor-agent.md","type":"page","name":"Cursor Agent CLI (headless)","description":"Binary is agent (the cursor-agent name is a back-compat alias). --force is mandatory for the CLI to actually apply file edits rather than only propose them; -p is the non-interactive/headless...","icon":"$undefined","url":"/docs/providers/cursor-agent","$ref":"providers/cursor-agent.md"}],"$id":"fallback:providers","$ref":{"folder":"providers","meta":"$undefined"},"icon":"$undefined"}]}},"tabs":[],"tabMode":"$undefined","sidebar":{"footer":"$L10"},"nav":{"title":"$L11"},"themeSwitch":{"enabled":false},"children":"$L12"}]}]]}]]}],"p":false,"v":null},"c":"$Q13"}]] 20:I[22530,["$2","$7","$8","$9","$a","$b","$c"],"ThemeToggle"] 22:I[97367,["$2"],"OutletBoundary"] 24:"/_next/static/chunks/2mrgr9w3cnjpj.js" 25:I[53266,["$2","$7","$8","$9","$a","$b","$c","$24"],"DocsPage"] 26:I[53266,["$2","$7","$8","$9","$a","$b","$c","$24"],"DocsTitle"] 27:I[53266,["$2","$7","$8","$9","$a","$b","$c","$24"],"DocsBody"] 28:I[27201,["$2"],"IconMark"] </article></body></html>

1:"$Sreact.fragment" 2:"/_next/static/chunks/1m2zgl4b4lvl0.js" 3:I[52488,["$2"],"ThemeSync"] 4:I[39756,["$2"],"default"] 5:I[37457,["$2"],"default"] 7:"/_next/static/chunks/2rp08ss88t7ft.js" 8:"/_next/static/chunks/3jqx1vbzj-pf-.js" 9:"/_next/static/chunks/2cspkb2ie5d02.js" a:"/_next/static/chunks/2g6d40dy4oi6g.js" b:"/_next/static/chunks/3qpt5v0fekgt8.js" c:"/_next/static/chunks/0a8prd-g29f0w.js" d:I[17751,["$2","$7","$8","$9","$a","$b","$c"],"RootProvider"] e:I[84694,["$2","$7","$8","$9","$a","$b","$c"],"default"] f:I[63199,["$2","$7","$8","$9","$a","$b","$c"],"LayoutBody"] 18:"ViewportBoundary" 19:I[97367,["$2"],"$18"] 1b:"MetadataBoundary" 1c:I[97367,["$2"],"$1b"] 1d:"$Sreact.suspense" 1f:I[68027,["$2"],"default",1] :HL["/_next/static/chunks/2f1h54i6zn087.css","style"] 15:X 16:[["children",{"s":"__PAGE__","h":512,"d":{"r":"$L17","p":false,"v":null}}]] 13:[["children",{"s":{"n":"slug","t":"oc","k":"usage/run-recipes/hub--narrow-terminals","s":[]},"h":512,"d":{"r":"$L14","p":false,"v":"$15"},"c":"$Q16"}]] 6:[["children",{"s":"docs","h":512,"d":{"r":["$","$1","c",{"children":[[["$","script","script-0",{"src":"/_next/static/chunks/2rp08ss88t7ft.js","async":true,"nonce":"$undefined"}],["$","script","script-1",{"src":"/_next/static/chunks/3jqx1vbzj-pf-.js","async":true,"nonce":"$undefined"}],["$","script","script-2",{"src":"/_next/static/chunks/2cspkb2ie5d02.js","async":true,"nonce":"$undefined"}],["$","script","script-3",{"src":"/_next/static/chunks/2g6d40dy4oi6g.js","async":true,"nonce":"$undefined"}],["$","script","script-4",{"src":"/_next/static/chunks/3qpt5v0fekgt8.js","async":true,"nonce":"$undefined"}],["$","script","script-5",{"src":"/_next/static/chunks/0a8prd-g29f0w.js","async":true,"nonce":"$undefined"}]],["$","$Ld",null,{"theme":{"enabled":false},"search":{"SearchDialog":"$e"},"children":[["$","a",null,{"className":"skip","href":"#docs-main","children":"Skip to content"}],["$","div",null,{"id":"docs-main","tabIndex":-1,"children":["$","$Lf",null,{"tree":{"type":"root","$ref":{"folder":"","meta":"meta.json"},"$id":"root","name":"Documentation","description":"$undefined","children":[{"$id":"index.mdx","type":"page","name":"Documentation","description":"Cyberine's developer documentation — install, dispatch, judge, ship.","icon":"$undefined","url":"/docs","$ref":"index.mdx"},{"$id":"getting-started.md","type":"page","name":"Getting started","description":"Install Cyberine, check that it can reach at least one worker, run a first task and read the verdict, then pin a repo and start on its task board.","icon":"$undefined","url":"/docs/getting-started","$ref":"getting-started.md"},{"$id":"beta.md","type":"page","name":"Beta guide","description":"What the closed Cyberine beta asks you to try, what it uses on your machine, how to upgrade, exactly what to delete to remove it, and where to report a problem.","icon":"$undefined","url":"/docs/beta","$ref":"beta.md"},{"$id":"concepts.md","type":"page","name":"Concepts","description":"The vocabulary of Cyberine in one place, defined once each - fleet, worker, judge, mode, seat, lane, loop worker, route, board, autopilot, draft, inbox and handoff.","icon":"$undefined","url":"/docs/concepts","$ref":"concepts.md"},{"$id":"features.md","type":"page","name":"Cyberine — Supported Features","description":"Cyberine is a vendor-independent, multi-provider AI agent orchestrator (ports-and-adapters): dispatch one task to N agents — raw-API loop workers and/or headless CLI coding agents — judge the...","icon":"$undefined","url":"/docs/features","$ref":"features.md"},{"type":"folder","name":"Usage","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"usage/index.md","type":"page","name":"Usage","description":"Every shipped Cyberine feature, its run recipe, and the full environment-variable table.","icon":"$undefined","url":"/docs/usage","$ref":"usage/index.md"},{"$id":"usage/install.md","type":"page","name":"Install","description":"The first-run happy path for a user with a logged-in claude (or codex, muse): install the CLI, check what it found, run one task, read the verdict, land the winner. Nothing else is needed for that...","icon":"$undefined","url":"/docs/usage/install","$ref":"usage/install.md"},{"$id":"usage/credentials-what-makes-each-adapter-runnable.md","type":"page","name":"Credentials (what makes each adapter runnable)","description":"buildDefaultAdapters() includes only the adapters whose credentials are present. Missing ones are noted and skipped - the run never crashes.","icon":"$undefined","url":"/docs/usage/credentials-what-makes-each-adapter-runnable","$ref":"usage/credentials-what-makes-each-adapter-runnable.md"},{"$id":"usage/configuration-file.md","type":"page","name":"Configuration file","description":"Cyberine supports layered JSON config files as an alternative to .env files. Precedence (highest first):","icon":"$undefined","url":"/docs/usage/configuration-file","$ref":"usage/configuration-file.md"},{"$id":"usage/cli-conventions--unknown-flags---dry-run-cost-scope-and-the---json-envelope-ks691-kstbv-kvw0q-kr3f0.md","type":"page","name":"CLI conventions — unknown flags, --dry-run, cost scope and the --json envelope (ks691, kstbv, kvw0q, kr3f0)","description":"Unknown flags are refused. init, keybindings, cost, credits, models, usage, status, cooldown, maintenance, enqueue, queue, fleet list, fleet init and plugin status check their arguments first...","icon":"$undefined","url":"/docs/usage/cli-conventions--unknown-flags---dry-run-cost-scope-and-the---json-envelope-ks691-kstbv-kvw0q-kr3f0","$ref":"usage/cli-conventions--unknown-flags---dry-run-cost-scope-and-the---json-envelope-ks691-kstbv-kvw0q-kr3f0.md"},{"type":"folder","name":"Run recipes","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"usage/run-recipes/index.md","type":"page","name":"Run recipes","description":"Run recipes: a Cyberine usage recipe.","icon":"$undefined","url":"/docs/usage/run-recipes","$ref":"usage/run-recipes/index.md"},{"$id":"usage/run-recipes/report-bugs-to-the-owning-repo--cyberine-issues.md","type":"page","name":"Report bugs to the owning repo — cyberine issues","description":"cyberine issues resolves an existing --about path with realpath and git, then appends the report to the owning repo. A Claude Code marketplace plugin cache path maps through...","icon":"$undefined","url":"/docs/usage/run-recipes/report-bugs-to-the-owning-repo--cyberine-issues","$ref":"usage/run-recipes/report-bugs-to-the-owning-repo--cyberine-issues.md"},{"$id":"usage/run-recipes/command-map-for-an-orchestrating-session--cyberine-guide.md","type":"page","name":"Command map for an orchestrating session — cyberine guide","description":"cyberine guide prints an intent-to-command cheatsheet for an ORCHESTRATING agent session (any harness with a shell, not only Claude Code) — the recipe to run first in a fresh session before...","icon":"$undefined","url":"/docs/usage/run-recipes/command-map-for-an-orchestrating-session--cyberine-guide","$ref":"usage/run-recipes/command-map-for-an-orchestrating-session--cyberine-guide.md"},{"$id":"usage/run-recipes/--help-on-any-subcommand-is-safe-kz4pg.md","type":"page","name":"--help on any subcommand is safe (kz4pg)","description":"--help (or -h) anywhere before a -- prints the subcommand's usage and exits 0. Most handlers print their own richer help; agents, chat, tui, setup, reap, maintenance, status and version used to...","icon":"$undefined","url":"/docs/usage/run-recipes/--help-on-any-subcommand-is-safe-kz4pg","$ref":"usage/run-recipes/--help-on-any-subcommand-is-safe-kz4pg.md"},{"$id":"usage/run-recipes/plugin-operating-rules--pluginreferencesmd.md","type":"page","name":"Plugin operating rules — plugin/references/*.md","description":"The plugin carries its own defaults for how a session works, so a Cyberine skill behaves the same on a machine whose Claude Code setup has no rules of its own (k6svf). Eight files under...","icon":"$undefined","url":"/docs/usage/run-recipes/plugin-operating-rules--pluginreferencesmd","$ref":"usage/run-recipes/plugin-operating-rules--pluginreferencesmd.md"},{"$id":"usage/run-recipes/plugin-cli-cheatsheet--pluginreferencescli-cheatsheetmd.md","type":"page","name":"Plugin CLI cheatsheet — plugin/references/cli-cheatsheet.md","description":"A Claude Code plugin skill cannot always shell out to cyberine guide before it needs the answer (a skill's own description budget is paid whether or not it ever runs a command), so the same...","icon":"$undefined","url":"/docs/usage/run-recipes/plugin-cli-cheatsheet--pluginreferencescli-cheatsheetmd","$ref":"usage/run-recipes/plugin-cli-cheatsheet--pluginreferencescli-cheatsheetmd.md"},{"$id":"usage/run-recipes/project-standard--cyberine-standard-and-cyberineplan---standard-kqjka.md","type":"page","name":"Project standard — cyberine standard and /cyberine:plan --standard (kqjka)","description":"Cyberine bundles a project standard: the areas a board is checked against so that an area nothing pulls in (observability, backup, CI/CD, release, legal, competitor watch) is still seen as...","icon":"$undefined","url":"/docs/usage/run-recipes/project-standard--cyberine-standard-and-cyberineplan---standard-kqjka","$ref":"usage/run-recipes/project-standard--cyberine-standard-and-cyberineplan---standard-kqjka.md"},{"$id":"usage/run-recipes/first-time-setup--cyberine-setup.md","type":"page","name":"First-time setup — cyberine setup","description":"Cyberine starts empty; cyberine setup is the guided onboarding (re-runnable, idempotent):","icon":"$undefined","url":"/docs/usage/run-recipes/first-time-setup--cyberine-setup","$ref":"usage/run-recipes/first-time-setup--cyberine-setup.md"},{"$id":"usage/run-recipes/project-level-setup--cyberine-init.md","type":"page","name":"Project-level setup — cyberine init","description":"cyberine init is the per-repo counterpart to cyberine setup. It detects which providers/CLIs are usable on the current machine, suggests a default fleet and mode, and writes...","icon":"$undefined","url":"/docs/usage/run-recipes/project-level-setup--cyberine-init","$ref":"usage/run-recipes/project-level-setup--cyberine-init.md"},{"$id":"usage/run-recipes/cyberine-sync-bring-a-repo-to-the-current-cyberine-kpwxe-2026-10-31.md","type":"page","name":"cyberine sync: bring a repo to the current Cyberine (kpwxe, 2026-10-31)","description":"Run recipe: cyberine sync in any repo that applied the Cyberine rules. It refreshes generated rule text (the Acceptance checks and Verification sections, when they still hold an older Cyberine...","icon":"$undefined","url":"/docs/usage/run-recipes/cyberine-sync-bring-a-repo-to-the-current-cyberine-kpwxe-2026-10-31","$ref":"usage/run-recipes/cyberine-sync-bring-a-repo-to-the-current-cyberine-kpwxe-2026-10-31.md"},{"$id":"usage/run-recipes/cyberine-rulings-keep-the-owner-rulings-file-from-growing-2026-11-06.md","type":"page","name":"cyberine rulings: keep the owner rulings file from growing (2026-11-06)","description":".claude/rules/cyberine-rulings.md loads into every session of a repo, and before this change it only grew: every owner decision stayed in it forever unless the owner reverted it. A test repo...","icon":"$undefined","url":"/docs/usage/run-recipes/cyberine-rulings-keep-the-owner-rulings-file-from-growing-2026-11-06","$ref":"usage/run-recipes/cyberine-rulings-keep-the-owner-rulings-file-from-growing-2026-11-06.md"},{"$id":"usage/run-recipes/cyberine-notice-scoped-owner-rulings-knowledge-and-plans-at-session-start-2026-11-10.md","type":"page","name":"cyberine notice: scoped owner rulings, knowledge and plans at session start (2026-11-10)","description":"A full rulings file loaded into every session and grew without bound, and Claude Code saves hook output over 10,000 characters to a file without telling anyone. Since k8380 (owner ruling D7-A) a...","icon":"$undefined","url":"/docs/usage/run-recipes/cyberine-notice-scoped-owner-rulings-knowledge-and-plans-at-session-start-2026-11-10","$ref":"usage/run-recipes/cyberine-notice-scoped-owner-rulings-knowledge-and-plans-at-session-start-2026-11-10.md"},{"$id":"usage/run-recipes/agent-policies-cyberine-init---policies.md","type":"page","name":"Agent policies (cyberine init --policies)","description":"cyberine init installs a public policy core: eight short, generic rules for how an agent works in any repo, one file each under .claude/rules/, written from the bundled templates in...","icon":"$undefined","url":"/docs/usage/run-recipes/agent-policies-cyberine-init---policies","$ref":"usage/run-recipes/agent-policies-cyberine-init---policies.md"},{"$id":"usage/run-recipes/smart-bare-cyberine.md","type":"page","name":"Smart bare cyberine","description":"Running cyberine with no subcommand is now state-aware:","icon":"$undefined","url":"/docs/usage/run-recipes/smart-bare-cyberine","$ref":"usage/run-recipes/smart-bare-cyberine.md"},{"$id":"usage/run-recipes/0-global-cyberine-command-cross-directory-run-monitor-à-la-claude-agents.md","type":"page","name":"0. Global cyberine command (cross-directory run monitor, à la claude agents)","description":"After npm link, the cyberine command works from ANY directory:","icon":"$undefined","url":"/docs/usage/run-recipes/0-global-cyberine-command-cross-directory-run-monitor-%C3%A0-la-claude-agents","$ref":"usage/run-recipes/0-global-cyberine-command-cross-directory-run-monitor-à-la-claude-agents.md"},{"$id":"usage/run-recipes/failover-preflight--cost-before-a-run-spends-anything-cyberine-route-cyberine-run.md","type":"page","name":"Failover preflight — cost before a run spends anything (cyberine route, cyberine run)","description":"Both cyberine route \"\" (dry run, nothing dispatched) and cyberine run \"\" (foreground and --bg, printed once to stderr before dispatch, no prompt) print one compact preflight block: the...","icon":"$undefined","url":"/docs/usage/run-recipes/failover-preflight--cost-before-a-run-spends-anything-cyberine-route-cyberine-run","$ref":"usage/run-recipes/failover-preflight--cost-before-a-run-spends-anything-cyberine-route-cyberine-run.md"},{"$id":"usage/run-recipes/machine-pressure-preflight.md","type":"page","name":"Machine pressure preflight","description":"Owner ruling 2026-10-29 (D1-A): Cyberine must not make a laptop that is already out of memory or disk worse. Before it spawns a NEW worker process, Cyberine reads a memory signal, free disk on the...","icon":"$undefined","url":"/docs/usage/run-recipes/machine-pressure-preflight","$ref":"usage/run-recipes/machine-pressure-preflight.md"},{"$id":"usage/run-recipes/command-code-over-acp-agent-client-protocol.md","type":"page","name":"Command Code over ACP (Agent Client Protocol)","description":"Run recipe:","icon":"$undefined","url":"/docs/usage/run-recipes/command-code-over-acp-agent-client-protocol","$ref":"usage/run-recipes/command-code-over-acp-agent-client-protocol.md"},{"$id":"usage/run-recipes/free-lane-out-of-credits--parked-not-failed.md","type":"page","name":"Free lane out of credits — parked, not failed","description":"A Command Code worker (loop-commandcode-free-build, loop-commandcode-build) that gets 400 You have insufficient credits to make this request is parked instead of failing as a plain error: the lane...","icon":"$undefined","url":"/docs/usage/run-recipes/free-lane-out-of-credits--parked-not-failed","$ref":"usage/run-recipes/free-lane-out-of-credits--parked-not-failed.md"},{"$id":"usage/run-recipes/wait-for--digest-a-background-run--cyberine-wait-cyberine-digest.md","type":"page","name":"Wait for / digest a background run — cyberine wait, cyberine digest","description":"A background run (cyberine run --bg, chat /run, npm run wf) prints a launch card that carries the run id (its first 8 characters). Two commands turn that id into a hands-off operator loop without...","icon":"$undefined","url":"/docs/usage/run-recipes/wait-for--digest-a-background-run--cyberine-wait-cyberine-digest","$ref":"usage/run-recipes/wait-for--digest-a-background-run--cyberine-wait-cyberine-digest.md"},{"$id":"usage/run-recipes/recurring-scheduled-dispatches--cyberine-cron.md","type":"page","name":"Recurring scheduled dispatches — cyberine cron","description":"Cyberine can re-dispatch a task on a schedule. Jobs are persisted in /cron.json (so they survive restarts) and are dispatched as detached background runs (cyberine run --bg) when a...","icon":"$undefined","url":"/docs/usage/run-recipes/recurring-scheduled-dispatches--cyberine-cron","$ref":"usage/run-recipes/recurring-scheduled-dispatches--cyberine-cron.md"},{"$id":"usage/run-recipes/batch-integrate-finished-runs--cyberine-land.md","type":"page","name":"Batch-integrate finished runs — cyberine land","description":"cyberine land waits for each run independently and commits each ready winner without waiting for another run's checks. Each winner is classified by its touched paths before publication. Except for...","icon":"$undefined","url":"/docs/usage/run-recipes/batch-integrate-finished-runs--cyberine-land","$ref":"usage/run-recipes/batch-integrate-finished-runs--cyberine-land.md"},{"$id":"usage/run-recipes/open-the-winner-as-a-draft-pull-request--cyberine-apply---pr--cyberine-run---pr.md","type":"page","name":"Open the winner as a draft pull request — cyberine apply --pr / cyberine run --pr","description":"cyberine apply lands a run's judged winner on your working tree. --pr sends the same winning diff to GitHub instead: it pushes a new branch and opens a DRAFT pull request whose body summarizes the...","icon":"$undefined","url":"/docs/usage/run-recipes/open-the-winner-as-a-draft-pull-request--cyberine-apply---pr--cyberine-run---pr","$ref":"usage/run-recipes/open-the-winner-as-a-draft-pull-request--cyberine-apply---pr--cyberine-run---pr.md"},{"$id":"usage/run-recipes/see-whats-runnable--dispatch-by-id--cyberine-next.md","type":"page","name":"See what's runnable + dispatch by id — cyberine next","description":"cyberine next derives the runnable / blocked / waiting sets from the project task manifest (.claude/TASKS.md or .local/tasks/ACTIVE.md) — the same deps + blocker logic cyberine tasks run uses, so...","icon":"$undefined","url":"/docs/usage/run-recipes/see-whats-runnable--dispatch-by-id--cyberine-next","$ref":"usage/run-recipes/see-whats-runnable--dispatch-by-id--cyberine-next.md"},{"$id":"usage/run-recipes/json-reads-of-the-board--status---json-and-next---json-ka7fx.md","type":"page","name":"JSON reads of the board — status --json and next --json (ka7fx)","description":"Two read-only views of the board print one JSON object on one stdout line, for a program that narrates the board (the cyberine-taskboard plugin, a script, a status bar) instead of parsing the text...","icon":"$undefined","url":"/docs/usage/run-recipes/json-reads-of-the-board--status---json-and-next---json-ka7fx","$ref":"usage/run-recipes/json-reads-of-the-board--status---json-and-next---json-ka7fx.md"},{"$id":"usage/run-recipes/list--switch-models--cyberine-models-cyberine-model.md","type":"page","name":"List + switch models — cyberine models, cyberine model","description":"The catalog is an embedded offline seed (src/core/modelCatalog.ts) and the sole source of truth — there is no runtime cache to fall behind it. cyberine model use writes nothing but the one MODEL...","icon":"$undefined","url":"/docs/usage/run-recipes/list--switch-models--cyberine-models-cyberine-model","$ref":"usage/run-recipes/list--switch-models--cyberine-models-cyberine-model.md"},{"$id":"usage/run-recipes/brainstorm-tasks-interactively--cyberine-chat.md","type":"page","name":"Brainstorm tasks interactively — cyberine chat","description":"cyberine run needs a task up front; cyberine chat is the interactive way to arrive at one. On a real TTY it opens a full-screen Ink TUI (a welcome banner with the model/provider/cwd + command panel, a","icon":"$undefined","url":"/docs/usage/run-recipes/brainstorm-tasks-interactively--cyberine-chat","$ref":"usage/run-recipes/brainstorm-tasks-interactively--cyberine-chat.md"},{"$id":"usage/run-recipes/task-workflow-inside-cyberine-chat--the-project-task-manifest.md","type":"page","name":"Task workflow inside cyberine chat — the project task manifest","description":"/run and the planner TASK: line dispatch ad-hoc text; the four slash commands below let the chat operate directly on the project's task manifest — the same .claude/TASKS.md / .codex/TASKS.md /...","icon":"$undefined","url":"/docs/usage/run-recipes/task-workflow-inside-cyberine-chat--the-project-task-manifest","$ref":"usage/run-recipes/task-workflow-inside-cyberine-chat--the-project-task-manifest.md"},{"$id":"usage/run-recipes/1-multi-adapter-demo-same-task---every-available-adapter-compare.md","type":"page","name":"1. Multi-adapter demo (same task -> every available adapter, compare)","description":"Dispatches one fixed task to all available adapters in parallel and prints a comparison table (provider | status | output | costUsd | ms) plus per-adapter detail. Proves every adapter returns the...","icon":"$undefined","url":"/docs/usage/run-recipes/1-multi-adapter-demo-same-task---every-available-adapter-compare","$ref":"usage/run-recipes/1-multi-adapter-demo-same-task---every-available-adapter-compare.md"},{"$id":"usage/run-recipes/2-one-off-task-from-the-cli.md","type":"page","name":"2. One-off task from the CLI","description":"Everything after -- is the task prompt. It is dispatched to all available adapters.","icon":"$undefined","url":"/docs/usage/run-recipes/2-one-off-task-from-the-cli","$ref":"usage/run-recipes/2-one-off-task-from-the-cli.md"},{"$id":"usage/run-recipes/3-switch-the-opencode-model--reasoning-effort-config-only.md","type":"page","name":"3. Switch the OpenCode model / reasoning effort (config only)","description":"Canonical ids only: deepseek/deepseek-v4-pro (default), deepseek/deepseek-v4-flash, moonshotai/kimi-k2.7-code, xiaomi/mimo-v2.6-pro. deepseek-chat / deepseek-reasoner were removed from DeepSeek's...","icon":"$undefined","url":"/docs/usage/run-recipes/3-switch-the-opencode-model--reasoning-effort-config-only","$ref":"usage/run-recipes/3-switch-the-opencode-model--reasoning-effort-config-only.md"},{"$id":"usage/run-recipes/4-switch-the-openrouter-loop-model-config-only.md","type":"page","name":"4. Switch the OpenRouter loop model (config only)","description":"Mind the date suffix — the two DeepSeek routes are inverted. On OpenRouter the DATED ids (deepseek/deepseek-v4-pro-0813, deepseek/deepseek-v4-flash-0731) are the current GAs and the UNDATED...","icon":"$undefined","url":"/docs/usage/run-recipes/4-switch-the-openrouter-loop-model-config-only","$ref":"usage/run-recipes/4-switch-the-openrouter-loop-model-config-only.md"},{"$id":"usage/run-recipes/5-full-gateway-path-anthropic--openai--any-provider-via-litellm.md","type":"page","name":"5. Full gateway path (Anthropic / OpenAI / any provider via LiteLLM)","description":"5. Full gateway path (Anthropic / OpenAI / any provider via LiteLLM): a Cyberine usage recipe.","icon":"$undefined","url":"/docs/usage/run-recipes/5-full-gateway-path-anthropic--openai--any-provider-via-litellm","$ref":"usage/run-recipes/5-full-gateway-path-anthropic--openai--any-provider-via-litellm.md"},{"$id":"usage/run-recipes/global-credentials-cyberine-from-any-directory.md","type":"page","name":"Global credentials (cyberine from any directory)","description":"Credentials load with precedence shell env > ~/.cyberine/.env (global, trusted) > ./.env (cwd); a key already set is never overridden. The global ~/.cyberine/.env is what lets the installed cyberine","icon":"$undefined","url":"/docs/usage/run-recipes/global-credentials-cyberine-from-any-directory","$ref":"usage/run-recipes/global-credentials-cyberine-from-any-directory.md"},{"$id":"usage/run-recipes/api-key-providers.md","type":"page","name":"API-key providers","description":"deepseek, moonshot, xiaomi, minimax, openai, google, xai, groq, meta, mistral, together, fireworks, azure, qwen, zhipu, perplexity, cerebras, nvidia, cohere, ollama-cloud, commandcode, openrouter,...","icon":"$undefined","url":"/docs/usage/run-recipes/api-key-providers","$ref":"usage/run-recipes/api-key-providers.md"},{"$id":"usage/run-recipes/github-copilot-removed-kn16x-2026-10-25.md","type":"page","name":"GitHub Copilot: removed (kn16x, 2026-10-25)","description":"Cyberine has no OAuth provider any more. The GitHub Copilot device flow and the loop-copilot-build worker were removed: they reused the VS Code Copilot OAuth client id to call the Copilot API outside","icon":"$undefined","url":"/docs/usage/run-recipes/github-copilot-removed-kn16x-2026-10-25","$ref":"usage/run-recipes/github-copilot-removed-kn16x-2026-10-25.md"},{"$id":"usage/run-recipes/minimax-mmx--text-worker--imagespeech-tools.md","type":"page","name":"MiniMax mmx — text worker + image/speech tools","description":"The MiniMax mmx CLI is two capabilities in Cyberine: a TEXT worker (mmx-text-build) and two multimedia IMPLEMENTER tools (mmximagegenerate, mmxspeechsynthesize). Auth is flexible — EITHER export...","icon":"$undefined","url":"/docs/usage/run-recipes/minimax-mmx--text-worker--imagespeech-tools","$ref":"usage/run-recipes/minimax-mmx--text-worker--imagespeech-tools.md"},{"$id":"usage/run-recipes/vision--multimodal-tasks.md","type":"page","name":"Vision / multimodal tasks","description":"Loop workers can forward image attachments to vision-capable models. Images are read from disk and sent as OpenAI-style imageurl content parts (base64 data URLs). The router automatically excludes...","icon":"$undefined","url":"/docs/usage/run-recipes/vision--multimodal-tasks","$ref":"usage/run-recipes/vision--multimodal-tasks.md"},{"$id":"usage/run-recipes/deepseek-request-isolation--user_id.md","type":"page","name":"DeepSeek request isolation — user_id","description":"Every DeepSeek-direct dispatch now carries DeepSeek's userid isolation field, which buys content-safety, KVCache, and scheduling isolation for sub-identities under one API account. On an account...","icon":"$undefined","url":"/docs/usage/run-recipes/deepseek-request-isolation--user_id","$ref":"usage/run-recipes/deepseek-request-isolation--user_id.md"},{"$id":"usage/run-recipes/the-opencode2-worker-opencode-20-preview.md","type":"page","name":"The opencode2 worker (OpenCode 2.0 preview)","description":"opencode2-deepseek is a build-fleet worker driving the OpenCode 2.0 preview binary. It is a SIBLING of opencode-deepseek, not a replacement — v1 and v2 are separate binaries, and the v1 one is...","icon":"$undefined","url":"/docs/usage/run-recipes/the-opencode2-worker-opencode-20-preview","$ref":"usage/run-recipes/the-opencode2-worker-opencode-20-preview.md"},{"$id":"usage/run-recipes/the-muse-worker-muse-code-meta.md","type":"page","name":"The muse worker (Muse Code, Meta)","description":"muse is a build-fleet CLI worker driving Meta's Muse Code agent. It is flat-rate against the CLI subscription (costModel: 'subscription', every catalog row reports cost: null), carries a ~1M-token...","icon":"$undefined","url":"/docs/usage/run-recipes/the-muse-worker-muse-code-meta","$ref":"usage/run-recipes/the-muse-worker-muse-code-meta.md"},{"$id":"usage/run-recipes/muse-via-the-raw-meta-model-api-loop-muse-build.md","type":"page","name":"Muse via the raw Meta Model API (loop-muse-build)","description":"A SECOND, unrelated way to reach the same models: a key-only LoopAgentAdapter against Meta's OpenAI-compatible endpoint, with no muse binary and no CLI login. It is registered but ships retired,...","icon":"$undefined","url":"/docs/usage/run-recipes/muse-via-the-raw-meta-model-api-loop-muse-build","$ref":"usage/run-recipes/muse-via-the-raw-meta-model-api-loop-muse-build.md"},{"$id":"usage/run-recipes/spawned-cli-workers-and-pwd.md","type":"page","name":"Spawned CLI workers and PWD","description":"childprocess.spawn() sets a child's working directory but copies PWD from the parent environment verbatim — a shell updates PWD on every cd, Node does not. So a CLI worker spawned into a git...","icon":"$undefined","url":"/docs/usage/run-recipes/spawned-cli-workers-and-pwd","$ref":"usage/run-recipes/spawned-cli-workers-and-pwd.md"},{"$id":"usage/run-recipes/retiring-a-worker--retired_workers.md","type":"page","name":"Retiring a worker — RETIRED_WORKERS","description":"A worker whose account is unfunded or ineligible is not the same as a worker with no credential, and Cyberine used to have no way to say so. An out-of-balance account still holds a VALID key, so...","icon":"$undefined","url":"/docs/usage/run-recipes/retiring-a-worker--retired_workers","$ref":"usage/run-recipes/retiring-a-worker--retired_workers.md"},{"$id":"usage/run-recipes/the-web-ui--cyberine-web-browser-fleet-monitor--chat.md","type":"page","name":"The web UI — cyberine web (browser fleet monitor + chat)","description":"cyberine web starts a dependency-free HTTP server (Node's built-in http, no framework) that serves a single-page browser UI mirroring the fleet TUI plus the chat planner. Same data as cyberine...","icon":"$undefined","url":"/docs/usage/run-recipes/the-web-ui--cyberine-web-browser-fleet-monitor--chat","$ref":"usage/run-recipes/the-web-ui--cyberine-web-browser-fleet-monitor--chat.md"},{"$id":"usage/run-recipes/6-workflows--live-fleet-view-the-orchestration-layer.md","type":"page","name":"6. Workflows + live fleet view (the orchestration layer)","description":"Run a multi-step workflow - a DAG of tasks where each step fans out to all available adapters, gets judged, and its winning output flows into dependent steps via {{stepId}} templating:","icon":"$undefined","url":"/docs/usage/run-recipes/6-workflows--live-fleet-view-the-orchestration-layer","$ref":"usage/run-recipes/6-workflows--live-fleet-view-the-orchestration-layer.md"},{"$id":"usage/run-recipes/swimlane-timeline-in-run-detail-kjvv7.md","type":"page","name":"Swimlane timeline in run detail (kjvv7)","description":"Run recipe: open a run with i in the hub (or cyberine tui), then press v to swap the step list for a timeline and v again to bring the steps back. The timeline is one lane per worker over the...","icon":"$undefined","url":"/docs/usage/run-recipes/swimlane-timeline-in-run-detail-kjvv7","$ref":"usage/run-recipes/swimlane-timeline-in-run-detail-kjvv7.md"},{"$id":"usage/run-recipes/meter-tracks-belong-to-their-own-meter-kw1e9.md","type":"page","name":"Meter tracks belong to their own meter (kw1e9)","description":"Two leftovers from the sub-cell meter work.","icon":"$undefined","url":"/docs/usage/run-recipes/meter-tracks-belong-to-their-own-meter-kw1e9","$ref":"usage/run-recipes/meter-tracks-belong-to-their-own-meter-kw1e9.md"},{"$id":"usage/run-recipes/the-sparkline-colours-by-value-kefc7.md","type":"page","name":"The sparkline colours by value (kefc7)","description":"The fleet's throughput sparkline maps each point to one of eight block levels. That level — how hot the point is — was computed and then thrown away, and the whole line rendered in a single...","icon":"$undefined","url":"/docs/usage/run-recipes/the-sparkline-colours-by-value-kefc7","$ref":"usage/run-recipes/the-sparkline-colours-by-value-kefc7.md"},{"$id":"usage/run-recipes/the-accent-budget--two-claims-per-screen-kpbk7.md","type":"page","name":"The accent budget — two claims per screen (kpbk7)","description":"The TUI carried 282 ACCENT references and a single hub screen painted twenty-plus orange spans: the header border and its breadcrumb, every key in the footer hint bar, each folder header, the \"N...","icon":"$undefined","url":"/docs/usage/run-recipes/the-accent-budget--two-claims-per-screen-kpbk7","$ref":"usage/run-recipes/the-accent-budget--two-claims-per-screen-kpbk7.md"},{"$id":"usage/run-recipes/divider-and-border-weight-k6shr.md","type":"page","name":"Divider and border weight (k6shr)","description":"Two related fixes to how a Divider (src/tui/components/Divider.tsx) sits next to the chrome around it, documented in full beside the accent budget and the spacing scale in src/tui/theme.ts:","icon":"$undefined","url":"/docs/usage/run-recipes/divider-and-border-weight-k6shr","$ref":"usage/run-recipes/divider-and-border-weight-k6shr.md"},{"$id":"usage/run-recipes/the-bold-budget--one-bold-per-region-km2w9.md","type":"page","name":"The bold budget — one bold per region (km2w9)","description":"The accent budget above rations hue; nothing rationed WEIGHT. A census of src/tui found bold set at 106 sites: every field label (task, verdict, chain, routing, live, tools, skills, name, detail,...","icon":"$undefined","url":"/docs/usage/run-recipes/the-bold-budget--one-bold-per-region-km2w9","$ref":"usage/run-recipes/the-bold-budget--one-bold-per-region-km2w9.md"},{"$id":"usage/run-recipes/the-border-vocabulary-is-corner-less-by-decision-kwk4f.md","type":"page","name":"The border vocabulary is corner-less, by decision (kwk4f)","description":"The TUI's chrome draws a top edge and a bottom edge and nothing else. There is exactly ONE border archetype, chromeBorder() in src/tui/theme.ts, no boxed variant, no left or right edge, and no...","icon":"$undefined","url":"/docs/usage/run-recipes/the-border-vocabulary-is-corner-less-by-decision-kwk4f","$ref":"usage/run-recipes/the-border-vocabulary-is-corner-less-by-decision-kwk4f.md"},{"$id":"usage/run-recipes/one-dot-separator-no-raw-colour-literal-no-doubled-hue-k1rse.md","type":"page","name":"One dot separator, no raw colour literal, no doubled hue (k1rse)","description":"A census (not an estimate) found the \"field · field · field\" dot separator (SYM.bullet, padded) rendered at THREE different widths across 31 pure separator sites in 11 files: two spaces before /...","icon":"$undefined","url":"/docs/usage/run-recipes/one-dot-separator-no-raw-colour-literal-no-doubled-hue-k1rse","$ref":"usage/run-recipes/one-dot-separator-no-raw-colour-literal-no-doubled-hue-k1rse.md"},{"$id":"usage/run-recipes/the-wide-hub-a-140-column-breakpoint-that-splits-list-and-preview-kxxph.md","type":"page","name":"The wide hub: a 140-column breakpoint that splits list and preview (kxxph)","description":"At 140 terminal columns and above, the cyberine agents hub stops being one column. The runs list becomes the left pane, the run preview becomes a full-height right pane, and a vertical rule sits...","icon":"$undefined","url":"/docs/usage/run-recipes/the-wide-hub-a-140-column-breakpoint-that-splits-list-and-preview-kxxph","$ref":"usage/run-recipes/the-wide-hub-a-140-column-breakpoint-that-splits-list-and-preview-kxxph.md"},{"$id":"usage/run-recipes/the-streaming-label-pulses-and-its-copy-escalates-kk92a.md","type":"page","name":"The streaming label pulses, and its copy escalates (kk92a)","description":"While a chat response is in flight the waiting label pulses between the muted tone and a slightly brighter one on a 2-second sine, after a 3-second warm-up. The warm-up is the load-bearing part: a...","icon":"$undefined","url":"/docs/usage/run-recipes/the-streaming-label-pulses-and-its-copy-escalates-kk92a","$ref":"usage/run-recipes/the-streaming-label-pulses-and-its-copy-escalates-kk92a.md"},{"$id":"usage/run-recipes/meters-render-at-sub-cell-resolution-k9y4k.md","type":"page","name":"Meters render at sub-cell resolution (k9y4k)","description":"Progress and context meters use the eighth-block ramp for the cell at the fill boundary, so a 16-cell bar shows 129 distinguishable levels instead of 17 — at the same width, because all eight...","icon":"$undefined","url":"/docs/usage/run-recipes/meters-render-at-sub-cell-resolution-k9y4k","$ref":"usage/run-recipes/meters-render-at-sub-cell-resolution-k9y4k.md"},{"$id":"usage/run-recipes/code-fences-use-four-tones-not-eight-kptd7.md","type":"page","name":"Code fences use four tones, not eight (kptd7)","description":"A rendered fence carried eight distinct colours. One was the raw ANSI name magenta, which bypassed the palette entirely and so rendered the same in every variant. One was the brand accent doing...","icon":"$undefined","url":"/docs/usage/run-recipes/code-fences-use-four-tones-not-eight-kptd7","$ref":"usage/run-recipes/code-fences-use-four-tones-not-eight-kptd7.md"},{"$id":"usage/run-recipes/a-judged-winner-looks-like-one-ksdrd.md","type":"page","name":"A judged winner looks like one (ksdrd)","description":"The three moments the product is actually about — the judge's pick, a DiffJudge apply-and-typecheck pass, a consensus verdict — used to render in the same tone as any successful thing on screen: a...","icon":"$undefined","url":"/docs/usage/run-recipes/a-judged-winner-looks-like-one-ksdrd","$ref":"usage/run-recipes/a-judged-winner-looks-like-one-ksdrd.md"},{"$id":"usage/run-recipes/the-fuzzy-picker-highlights-what-it-matched-kenqx.md","type":"page","name":"The fuzzy picker highlights what it matched (kenqx)","description":"Typing in any fuzzy picker now paints the characters that matched, the way fzf, helix and telescope do. The scorer already knew them — it tracks the position of every hit to compute its...","icon":"$undefined","url":"/docs/usage/run-recipes/the-fuzzy-picker-highlights-what-it-matched-kenqx","$ref":"usage/run-recipes/the-fuzzy-picker-highlights-what-it-matched-kenqx.md"},{"$id":"usage/run-recipes/the-stall-indicator-stays-inside-the-palette-kfapa.md","type":"page","name":"The stall indicator stays inside the palette (kfapa)","description":"While a reply is streaming, the input frame drifts from the brand accent through the warning tone to the error tone as the stream stays silent, reaching its end state after 8 seconds — the same...","icon":"$undefined","url":"/docs/usage/run-recipes/the-stall-indicator-stays-inside-the-palette-kfapa","$ref":"usage/run-recipes/the-stall-indicator-stays-inside-the-palette-kfapa.md"},{"$id":"usage/run-recipes/one-state-grammar-across-every-pane-kctsa.md","type":"page","name":"One state grammar across every pane (kctsa)","description":"Run and worker state colours come from stateColor() everywhere, so every surface teaches the same language: a blue spinner means working, a blue check means done.","icon":"$undefined","url":"/docs/usage/run-recipes/one-state-grammar-across-every-pane-kctsa","$ref":"usage/run-recipes/one-state-grammar-across-every-pane-kctsa.md"},{"$id":"usage/run-recipes/the-chat-repl-is-a-bottom-anchored-column-kntve.md","type":"page","name":"The chat REPL is a bottom-anchored column (kntve)","description":"The conversation grows UPWARD off the input, the way every good REPL does. Before this the transcript was a fixed-height box with no justifyContent, so on a 50-row terminal two messages sat at the...","icon":"$undefined","url":"/docs/usage/run-recipes/the-chat-repl-is-a-bottom-anchored-column-kntve","$ref":"usage/run-recipes/the-chat-repl-is-a-bottom-anchored-column-kntve.md"},{"$id":"usage/run-recipes/diffs-are-filled-not-coloured-k7pc0.md","type":"page","name":"Diffs are filled, not coloured (k7pc0)","description":"Added and removed lines in the diff viewer render as a line BACKGROUND with readable text on top, the way delta and git render a diff — not as coloured glyphs.","icon":"$undefined","url":"/docs/usage/run-recipes/diffs-are-filled-not-coloured-k7pc0","$ref":"usage/run-recipes/diffs-are-filled-not-coloured-k7pc0.md"},{"$id":"usage/run-recipes/wide-terminals--the-hub-measures-rows-not-the-screen-k09p3.md","type":"page","name":"Wide terminals — the hub measures rows, not the screen (k09p3)","description":"Tabular content tolerates far more width than prose, but not unbounded width. At 200 columns one hub run summary occupied about 146 cells on a single line, the run preview truncated its wrapped...","icon":"$undefined","url":"/docs/usage/run-recipes/wide-terminals--the-hub-measures-rows-not-the-screen-k09p3","$ref":"usage/run-recipes/wide-terminals--the-hub-measures-rows-not-the-screen-k09p3.md"},{"$id":"usage/run-recipes/keybindings--remap-fleet--chat-actions-cyberinekeybindingsjson-t535.md","type":"page","name":"Keybindings — remap fleet + chat actions (~/.cyberine/keybindings.json, T535)","description":"Both Ink TUIs — the fleet hub (cyberine agents / npm run tui, namespace fleet.) and the chat REPL (cyberine chat, namespace chat.) — read remaps from ONE file: $CYBERINEKEYBINDINGS or...","icon":"$undefined","url":"/docs/usage/run-recipes/keybindings--remap-fleet--chat-actions-cyberinekeybindingsjson-t535","$ref":"usage/run-recipes/keybindings--remap-fleet--chat-actions-cyberinekeybindingsjson-t535.md"},{"$id":"usage/run-recipes/fleet-hub--blocks-inspector-and-motion.md","type":"page","name":"Fleet hub — blocks, inspector and motion","description":"cyberine agents (and cyberine tui) open onto a triage view (kp3fe): what needs you first, the one selected item in an inspector, and the finished archive closed behind a drawer. There is no...","icon":"$undefined","url":"/docs/usage/run-recipes/fleet-hub--blocks-inspector-and-motion","$ref":"usage/run-recipes/fleet-hub--blocks-inspector-and-motion.md"},{"$id":"usage/run-recipes/hub--run-titles-filter-scrolling.md","type":"page","name":"Hub — run titles, filter, scrolling","description":"Every row leads with a short title (at most 48 cells) for what the item is actually doing. For a run the engine derives it when the run is created and stores it as title in run.json: injected...","icon":"$undefined","url":"/docs/usage/run-recipes/hub--run-titles-filter-scrolling","$ref":"usage/run-recipes/hub--run-titles-filter-scrolling.md"},{"$id":"usage/run-recipes/hub--narrow-terminals.md","type":"page","name":"Hub — narrow terminals","description":"Below 80 columns the hub drops the agents and reason columns and caps the title at 24 cells; the age always stays. The hint row keeps only the hints that fit on ONE line, most essential first, and...","icon":"$undefined","url":"/docs/usage/run-recipes/hub--narrow-terminals","$ref":"usage/run-recipes/hub--narrow-terminals.md"},{"$id":"usage/run-recipes/hub--reply-from-the-peek-space.md","type":"page","name":"Hub — reply from the peek (Space)","description":"The former quick-reply panel and the separate agents view (c) are gone: the hub lists runs and chats together, and Space opens the inspector on the selected row with a reply input (see \"Fleet hub\").","icon":"$undefined","url":"/docs/usage/run-recipes/hub--reply-from-the-peek-space","$ref":"usage/run-recipes/hub--reply-from-the-peek-space.md"},{"$id":"usage/run-recipes/hub--opening-runs-and-chats.md","type":"page","name":"Hub — opening runs and chats","description":"⏎ opens the right thing for the row kind:","icon":"$undefined","url":"/docs/usage/run-recipes/hub--opening-runs-and-chats","$ref":"usage/run-recipes/hub--opening-runs-and-chats.md"},{"$id":"usage/run-recipes/hub--folders-collapse-and-hide.md","type":"page","name":"Hub — folders, collapse and hide","description":"Ctrl+S until the strip reads group: folder groups the hub by project directory. Every folder header shows the folder name and an item count (~/work/api (7)), plus pinned when the folder is pinned....","icon":"$undefined","url":"/docs/usage/run-recipes/hub--folders-collapse-and-hide","$ref":"usage/run-recipes/hub--folders-collapse-and-hide.md"},{"$id":"usage/run-recipes/run-detail--task-panel-prompt-verdict-diff-race-timing.md","type":"page","name":"Run detail — task panel, prompt, verdict, diff, race timing","description":"Inspecting a run shows a task line under the run id (the same summary as the list), and each step shows a verdict card — the judge method, the winning worker, and the agreement % when available —...","icon":"$undefined","url":"/docs/usage/run-recipes/run-detail--task-panel-prompt-verdict-diff-race-timing","$ref":"usage/run-recipes/run-detail--task-panel-prompt-verdict-diff-race-timing.md"},{"$id":"usage/run-recipes/fleet-list--rate-column-retired.md","type":"page","name":"Fleet list — RATE column (retired)","description":"The RATE sparkline column belonged to the old columned fleet list and is gone from the direction-A hub, which shows one line per item with no throughput column. The worker peek header still draws...","icon":"$undefined","url":"/docs/usage/run-recipes/fleet-list--rate-column-retired","$ref":"usage/run-recipes/fleet-list--rate-column-retired.md"},{"$id":"usage/run-recipes/tui--flicker-free-rendering-synchronized-output.md","type":"page","name":"TUI — flicker-free rendering (synchronized output)","description":"Ink repaints the whole frame on every animation tick (every 120ms while runs are live), which on most terminals causes visible full-screen flicker. Every Ink frame write is wrapped in DEC 2026...","icon":"$undefined","url":"/docs/usage/run-recipes/tui--flicker-free-rendering-synchronized-output","$ref":"usage/run-recipes/tui--flicker-free-rendering-synchronized-output.md"},{"$id":"usage/run-recipes/tuichat--local-notifications-t534.md","type":"page","name":"TUI/chat — local notifications (T534)","description":"A shared notify channel (src/tui/notify.ts) fires a local notification — never a network call, never OAuth — from these triggers:","icon":"$undefined","url":"/docs/usage/run-recipes/tuichat--local-notifications-t534","$ref":"usage/run-recipes/tuichat--local-notifications-t534.md"},{"$id":"usage/run-recipes/tuichat--terminal-tab-status-osc-7501-k33z2.md","type":"page","name":"TUI/chat — terminal tab status, OSC 7501 (k33z2)","description":"cyberine chat and the fleet hub (cyberine agents / npm run tui) publish their state through the Program Status Protocol, the escape sequence Claude Code 2.1.295 uses so a terminal can show a tab...","icon":"$undefined","url":"/docs/usage/run-recipes/tuichat--terminal-tab-status-osc-7501-k33z2","$ref":"usage/run-recipes/tuichat--terminal-tab-status-osc-7501-k33z2.md"},{"$id":"usage/run-recipes/tuichat--update-notifier-t533.md","type":"page","name":"TUI/chat — update notifier (T533)","description":"On cyberine agents / npm run tui and cyberine chat startup, a background check (src/core/updateCheck.ts) compares the running version against the latest available and, if newer, shows one muted...","icon":"$undefined","url":"/docs/usage/run-recipes/tuichat--update-notifier-t533","$ref":"usage/run-recipes/tuichat--update-notifier-t533.md"},{"$id":"usage/run-recipes/tui--worker-toast-notifications.md","type":"page","name":"TUI — worker toast notifications","description":"A small stack of status lines appears between the content area and the key-hint footer whenever a worker finishes or fails — so a state change is noticed without watching rows.","icon":"$undefined","url":"/docs/usage/run-recipes/tui--worker-toast-notifications","$ref":"usage/run-recipes/tui--worker-toast-notifications.md"},{"$id":"usage/run-recipes/tui-end-to-end-cases--drive-chat-and-the-fleet-hub-through-a-real-terminal-npm-run-tuie2e.md","type":"page","name":"TUI end-to-end cases — drive chat and the fleet hub through a real terminal (npm run tui:e2e)","description":"Starts cyberine chat, cyberine agents and cyberine tui from this source tree inside a private tmux server, sends real keys, waits for what a person would see, and runs a UX invariant scan on every...","icon":"$undefined","url":"/docs/usage/run-recipes/tui-end-to-end-cases--drive-chat-and-the-fleet-hub-through-a-real-terminal-npm-run-tuie2e","$ref":"usage/run-recipes/tui-end-to-end-cases--drive-chat-and-the-fleet-hub-through-a-real-terminal-npm-run-tuie2e.md"},{"$id":"usage/run-recipes/workflow-monitor--live-phaseagent-drill-down-w.md","type":"page","name":"Workflow Monitor — live phase/agent drill-down (w)","description":"Press w on a selected run (from the fleet list or from a run's detail screen) to open the Workflow Monitor for that run. It replaces the current screen; Esc returns you to where you came from...","icon":"$undefined","url":"/docs/usage/run-recipes/workflow-monitor--live-phaseagent-drill-down-w","$ref":"usage/run-recipes/workflow-monitor--live-phaseagent-drill-down-w.md"},{"$id":"usage/run-recipes/stopping-a-run-or-worker.md","type":"page","name":"Stopping a run or worker","description":"s (stop run) / x (stop worker) work cross-process: the TUI and the engine run in separate processes and share only the filesystem, so stop is a control file the engine polls.","icon":"$undefined","url":"/docs/usage/run-recipes/stopping-a-run-or-worker","$ref":"usage/run-recipes/stopping-a-run-or-worker.md"},{"$id":"usage/run-recipes/foreground-run-wall-clock-timeout--cyberine-run---timeout-sec.md","type":"page","name":"Foreground run wall-clock timeout — cyberine run --timeout ","description":"--timeout adds an inline wall-clock limit to a foreground cyberine run. When the run has not finished within the given number of seconds, the CLI drops a stop control message carrying...","icon":"$undefined","url":"/docs/usage/run-recipes/foreground-run-wall-clock-timeout--cyberine-run---timeout-sec","$ref":"usage/run-recipes/foreground-run-wall-clock-timeout--cyberine-run---timeout-sec.md"},{"$id":"usage/run-recipes/live-progress-output--streaming-per-worker-status.md","type":"page","name":"Live progress output — streaming per-worker status","description":"cyberine run and npm run cli now emit live per-worker status lines to stderr as adapters work, rather than surfacing a single blob at the end. The per-worker lines appear as they arrive; the final...","icon":"$undefined","url":"/docs/usage/run-recipes/live-progress-output--streaming-per-worker-status","$ref":"usage/run-recipes/live-progress-output--streaming-per-worker-status.md"},{"$id":"usage/run-recipes/auto-routed-implementer-fleet-on-the-foreground-run.md","type":"page","name":"Auto-routed implementer fleet on the foreground run","description":"A bare cyberine run \"\" (no --fleet, no --cheap, no CYBERINEBUILDFLEET) used to run only the read-only compare pool and exit 0 with state: done and no diff. The foreground path now...","icon":"$undefined","url":"/docs/usage/run-recipes/auto-routed-implementer-fleet-on-the-foreground-run","$ref":"usage/run-recipes/auto-routed-implementer-fleet-on-the-foreground-run.md"},{"$id":"usage/run-recipes/cyberine-run-refuses-a-repository-with-no-commit-yet-k4zpd.md","type":"page","name":"cyberine run refuses a repository with no commit yet (k4zpd)","description":"Every implementer works in a detached git worktree off HEAD, so a repo that has git init but no commit cannot dispatch anything: before 2026-09-12 the run got a run id, reserved spend, and then...","icon":"$undefined","url":"/docs/usage/run-recipes/cyberine-run-refuses-a-repository-with-no-commit-yet-k4zpd","$ref":"usage/run-recipes/cyberine-run-refuses-a-repository-with-no-commit-yet-k4zpd.md"},{"$id":"usage/run-recipes/foreground-cyberine-run-with-a-build-fleet-runs-through-the-engine-isolated-worktrees.md","type":"page","name":"Foreground cyberine run with a build fleet runs through the engine (isolated worktrees)","description":"When a build fleet is in effect (routed, --fleet, --cheap, or CYBERINEBUILDFLEET) AND the cwd is inside a git repository, the foreground cyberine run no longer dispatches the raw compare pool...","icon":"$undefined","url":"/docs/usage/run-recipes/foreground-cyberine-run-with-a-build-fleet-runs-through-the-engine-isolated-worktrees","$ref":"usage/run-recipes/foreground-cyberine-run-with-a-build-fleet-runs-through-the-engine-isolated-worktrees.md"},{"$id":"usage/run-recipes/independent-answers-from-a-pinned-fleet--cyberine-run---compare-kmrcz.md","type":"page","name":"Independent answers from a pinned fleet — cyberine run --compare (kmrcz)","description":"A build fleet inside a git repository always means the implement-task chain above, whatever the prompt says: the design winner is pasted into every implementer's prompt, so the second worker reads...","icon":"$undefined","url":"/docs/usage/run-recipes/independent-answers-from-a-pinned-fleet--cyberine-run---compare-kmrcz","$ref":"usage/run-recipes/independent-answers-from-a-pinned-fleet--cyberine-run---compare-kmrcz.md"},{"$id":"usage/run-recipes/a-mixed-fleet-dispatches-to-every-active-worker-kw920.md","type":"page","name":"A mixed fleet dispatches to every active worker (kw920)","description":"With no build fleet pinned, every default dispatch now runs on the whole active fleet: the always-on key loop workers (loop-deepseek-direct and the other -direct workers enabled in fleet.json)...","icon":"$undefined","url":"/docs/usage/run-recipes/a-mixed-fleet-dispatches-to-every-active-worker-kw920","$ref":"usage/run-recipes/a-mixed-fleet-dispatches-to-every-active-worker-kw920.md"},{"$id":"usage/run-recipes/blind-candidate-ids-in-the-llm-and-harvest-judges.md","type":"page","name":"Blind candidate ids in the LLM and harvest judges","description":"llm-judge (and DiffJudge's tie-break, which runs it) shows candidates as anonymous ids A, B, C in a deterministic shuffle seeded by the task id and prompt, never as the worker's agent id or in...","icon":"$undefined","url":"/docs/usage/run-recipes/blind-candidate-ids-in-the-llm-and-harvest-judges","$ref":"usage/run-recipes/blind-candidate-ids-in-the-llm-and-harvest-judges.md"},{"$id":"usage/run-recipes/judge-backend-fallback-and-unjudged-runs.md","type":"page","name":"Judge backend fallback and unjudged runs","description":"The LLM judges (llm-judge, adversarial), the research ClaimVerifier and the SynthesizerReducer use every configured backend in precedence order. Without routing.judges in fleet.json the candidates...","icon":"$undefined","url":"/docs/usage/run-recipes/judge-backend-fallback-and-unjudged-runs","$ref":"usage/run-recipes/judge-backend-fallback-and-unjudged-runs.md"},{"$id":"usage/run-recipes/stale--zombie-runs.md","type":"page","name":"Stale / zombie runs","description":"If the engine process is killed mid-run (crash, Ctrl-C, kill), its run.json is left stuck in running with nothing driving it. The TUI detects this: the engine records its PID (enginePid) when it...","icon":"$undefined","url":"/docs/usage/run-recipes/stale--zombie-runs","$ref":"usage/run-recipes/stale--zombie-runs.md"},{"$id":"usage/run-recipes/cli-reaper--cyberine-reap---dry-run.md","type":"page","name":"CLI reaper — cyberine reap [--dry-run]","description":"cyberine reap is the command-line companion to the TUI's stale-run detection. It sweeps the global registry for runs whose engine subprocess is alive but has produced no events.jsonl progress for...","icon":"$undefined","url":"/docs/usage/run-recipes/cli-reaper--cyberine-reap---dry-run","$ref":"usage/run-recipes/cli-reaper--cyberine-reap---dry-run.md"},{"$id":"usage/run-recipes/always-on-operations--graceful-drain-health-queue-intake-maintenance.md","type":"page","name":"Always-on operations — graceful drain, health, queue intake, maintenance","description":"The API-first 24/7 model keeps cyberine web supervised and pokes it (or the CLI) with thin external triggers instead of running a fat internal daemon. The pieces below harden that path.","icon":"$undefined","url":"/docs/usage/run-recipes/always-on-operations--graceful-drain-health-queue-intake-maintenance","$ref":"usage/run-recipes/always-on-operations--graceful-drain-health-queue-intake-maintenance.md"},{"$id":"usage/run-recipes/standby-worker-pool-opt-in.md","type":"page","name":"Standby worker pool (opt-in)","description":"By default, every cyberine chat /run (and cyberine run --bg) cold-spawns a fresh detached child, paying the Node+tsx startup cost each time. The standby pool pre-warms N named worker slots so a...","icon":"$undefined","url":"/docs/usage/run-recipes/standby-worker-pool-opt-in","$ref":"usage/run-recipes/standby-worker-pool-opt-in.md"},{"$id":"usage/run-recipes/steering-a-running-loop-worker-between-tool-rounds.md","type":"page","name":"Steering a running loop worker (between tool rounds)","description":"control.json carries one of two actions — stop (above) or steer. A steer message injects free-text operator guidance into a running worker between tool rounds for loop workers, or at the next step...","icon":"$undefined","url":"/docs/usage/run-recipes/steering-a-running-loop-worker-between-tool-rounds","$ref":"usage/run-recipes/steering-a-running-loop-worker-between-tool-rounds.md"},{"$id":"usage/run-recipes/7-implementer-fleet-do-a-real-task-multiple-providers-compete-judge-picks-the-best-diff.md","type":"page","name":"7. Implementer fleet: do a real task (multiple providers compete, judge picks the best diff)","description":"Beyond Q&A/design, Cyberine can run a real file-changing task as an implementer fleet: several vendor coding agents each implement the same task in their OWN isolated git worktree, and a DiffJudge","icon":"$undefined","url":"/docs/usage/run-recipes/7-implementer-fleet-do-a-real-task-multiple-providers-compete-judge-picks-the-best-diff","$ref":"usage/run-recipes/7-implementer-fleet-do-a-real-task-multiple-providers-compete-judge-picks-the-best-diff.md"},{"$id":"usage/run-recipes/which-verdict-wins-a-step-k0fxw.md","type":"page","name":"Which verdict wins a step (k0fxw)","description":"Every judge in a step's panel returns a verdict; the step records ONE of them as verdict. ConsensusJudge wins only when a real majority of candidates produced the same answer (agreement above 0.5,...","icon":"$undefined","url":"/docs/usage/run-recipes/which-verdict-wins-a-step-k0fxw","$ref":"usage/run-recipes/which-verdict-wins-a-step-k0fxw.md"},{"$id":"usage/run-recipes/judge-phase-is-fail-bounded-t551.md","type":"page","name":"Judge phase is fail-bounded (T551)","description":"Every judge (ConsensusJudge, LlmJudge, AdversarialJudge, DiffJudge, BehaviorJudge, SwarmMergeJudge/LaneMergeJudge) is raced against CYBERINEJUDGETIMEOUTMS (default 60000ms) in...","icon":"$undefined","url":"/docs/usage/run-recipes/judge-phase-is-fail-bounded-t551","$ref":"usage/run-recipes/judge-phase-is-fail-bounded-t551.md"},{"$id":"usage/run-recipes/adversarialjudge-rationale-carries-the-skeptics-reasons-kxqep.md","type":"page","name":"AdversarialJudge rationale carries the skeptics' reasons (kxqep)","description":"AdversarialJudge already asked each skeptic for {\"refuted\": true|false, \"reason\": \"...\"}; the winning candidate's verdict.rationale now names the refuting skeptics' reasons instead of just the...","icon":"$undefined","url":"/docs/usage/run-recipes/adversarialjudge-rationale-carries-the-skeptics-reasons-kxqep","$ref":"usage/run-recipes/adversarialjudge-rationale-carries-the-skeptics-reasons-kxqep.md"},{"$id":"usage/run-recipes/auto-evolve-cyberine-authors-and-implements-its-own-task-mine---implement---gate.md","type":"page","name":"Auto-evolve: Cyberine authors and implements its own task (mine -> implement -> gate)","description":"The self-improvement loop with NO human-written task: a read-only worker surveys the repo and AUTHORS one small improvement spec (strict TASK/FILES/WHY/SPEC/ACCEPTANCE format, hard-bounded: 1-3...","icon":"$undefined","url":"/docs/usage/run-recipes/auto-evolve-cyberine-authors-and-implements-its-own-task-mine---implement---gate","$ref":"usage/run-recipes/auto-evolve-cyberine-authors-and-implements-its-own-task-mine---implement---gate.md"},{"$id":"usage/run-recipes/lspcompiler-diagnostic-feedback-loop-t436.md","type":"page","name":"LSP/compiler diagnostic feedback loop (T436)","description":"Loop-adapter workers (loop--build) can now self-correct type errors within a single run, matching opencode's built-in edit-then-check loop. After each tool round that mutates a file the adapter...","icon":"$undefined","url":"/docs/usage/run-recipes/lspcompiler-diagnostic-feedback-loop-t436","$ref":"usage/run-recipes/lspcompiler-diagnostic-feedback-loop-t436.md"},{"$id":"usage/run-recipes/verify-error-steering--test-runner-output-as-navigation-index-t447.md","type":"page","name":"Verify-error steering — test-runner output as navigation index (T447)","description":"T447 extends the T436 feedback loop to test-runner and stack-trace output (the STEER act). When a loop-adapter worker produces verify output — a codeexecution snippet that shells out to npm test,...","icon":"$undefined","url":"/docs/usage/run-recipes/verify-error-steering--test-runner-output-as-navigation-index-t447","$ref":"usage/run-recipes/verify-error-steering--test-runner-output-as-navigation-index-t447.md"},{"$id":"usage/run-recipes/long-term-memory--episodic-records-retrieval--injection-waves-1-4.md","type":"page","name":"Long-term memory — episodic records, retrieval + injection (waves 1-4)","description":"Every run that reaches a terminal state (done / error / cancelled) automatically writes an episodic memory record — no setup needed:","icon":"$undefined","url":"/docs/usage/run-recipes/long-term-memory--episodic-records-retrieval--injection-waves-1-4","$ref":"usage/run-recipes/long-term-memory--episodic-records-retrieval--injection-waves-1-4.md"},{"$id":"usage/run-recipes/rules--project--global-conventions-like-claude-code-rules.md","type":"page","name":"Rules — project + global conventions (like Claude Code rules)","description":"Drop markdown rule files and every dispatched task gets them injected as a RULES preamble (instructions), ahead of the MEMORY preamble (context) and the task — so cyberine follows your conventions...","icon":"$undefined","url":"/docs/usage/run-recipes/rules--project--global-conventions-like-claude-code-rules","$ref":"usage/run-recipes/rules--project--global-conventions-like-claude-code-rules.md"},{"$id":"usage/run-recipes/ecosystem-instruction-files-agentsmd--claudemd.md","type":"page","name":"Ecosystem instruction files (AGENTS.md / CLAUDE.md)","description":"Loop workers (raw-API, not CLI) automatically discover and read AGENTS.md / CLAUDE.md from the target repo and inject them as system-message context before the task. This gives every loop worker...","icon":"$undefined","url":"/docs/usage/run-recipes/ecosystem-instruction-files-agentsmd--claudemd","$ref":"usage/run-recipes/ecosystem-instruction-files-agentsmd--claudemd.md"},{"$id":"usage/run-recipes/lsp-worker-context-lsp_query-tool-t312.md","type":"page","name":"LSP worker context (lsp_query tool, T312)","description":"Loop workers can query live symbol, definition, reference, and diagnostic information from a Language Server Protocol (LSP) server. A shared warm connection is reused across all workers in a run —...","icon":"$undefined","url":"/docs/usage/run-recipes/lsp-worker-context-lsp_query-tool-t312","$ref":"usage/run-recipes/lsp-worker-context-lsp_query-tool-t312.md"},{"$id":"usage/run-recipes/anthropic-chat-prompt-caching-kv0qh.md","type":"page","name":"Anthropic chat prompt caching (kv0qh)","description":"The chat planner puts cachecontrol breakpoints on the trailing system block and the last tool definition of every request whose provider routes to an Anthropic model (the same policy as the...","icon":"$undefined","url":"/docs/usage/run-recipes/anthropic-chat-prompt-caching-kv0qh","$ref":"usage/run-recipes/anthropic-chat-prompt-caching-kv0qh.md"},{"$id":"usage/run-recipes/persona--chat-planner-voice-cyberine_persona.md","type":"page","name":"Persona — chat planner voice (CYBERINE_PERSONA)","description":"The chat planner (cyberine chat, cyberine web) opens every session with a persona preamble prepended above its operational instructions. The persona controls tone and voice; operational behavior...","icon":"$undefined","url":"/docs/usage/run-recipes/persona--chat-planner-voice-cyberine_persona","$ref":"usage/run-recipes/persona--chat-planner-voice-cyberine_persona.md"},{"$id":"usage/run-recipes/output-styles--output-style-persona-overlay-t539.md","type":"page","name":"Output styles — /output-style persona overlay (T539)","description":"/output-style is a SEPARATE toggle from /style (get/set the chat TRANSCRIPT verbosity — compact/verbose/json, run /help inside cyberine chat for the full command list) and from CYBERINEPERSONA...","icon":"$undefined","url":"/docs/usage/run-recipes/output-styles--output-style-persona-overlay-t539","$ref":"usage/run-recipes/output-styles--output-style-persona-overlay-t539.md"},{"$id":"usage/run-recipes/confluence-ring--welcome-screen-brand-mark.md","type":"page","name":"Confluence ring — welcome-screen brand mark","description":"The cyberine chat welcome header shows the Cyberine brand mark (the fleet hub dropped its overview splash and the mark with it in direction A, 2026-10-03), the \"confluence ring\": an open ring for...","icon":"$undefined","url":"/docs/usage/run-recipes/confluence-ring--welcome-screen-brand-mark","$ref":"usage/run-recipes/confluence-ring--welcome-screen-brand-mark.md"},{"$id":"usage/run-recipes/skills--claude-code-compatible-skillmd.md","type":"page","name":"Skills — Claude-Code-compatible SKILL.md","description":"cyberine discovers SKILL.md files in the SAME format Claude Code uses, so existing skills and templates transfer unchanged (unknown frontmatter keys like version / tools are ignored):","icon":"$undefined","url":"/docs/usage/run-recipes/skills--claude-code-compatible-skillmd","$ref":"usage/run-recipes/skills--claude-code-compatible-skillmd.md"},{"$id":"usage/run-recipes/cyberine-mcp-install--register-the-cyberine-mcp-server-with-a-harness.md","type":"page","name":"cyberine mcp install — register the cyberine MCP server with a harness","description":"Companion to cyberine skill install: instead of copying files, it registers cyberine (src/mcp.ts — the orchestrator's own MCP server, exposing orchestrate, research, the bus tools and the...","icon":"$undefined","url":"/docs/usage/run-recipes/cyberine-mcp-install--register-the-cyberine-mcp-server-with-a-harness","$ref":"usage/run-recipes/cyberine-mcp-install--register-the-cyberine-mcp-server-with-a-harness.md"},{"$id":"usage/run-recipes/backlog-tasks--whats-pending-beyond-run-memory.md","type":"page","name":"Backlog tasks — what's pending beyond run-memory","description":"cyberine tasks dogfood surfaces OPEN items from the project's .local/dogfood/backlog-.md files, so \"what's pending\" is not conflated with \"no recent runs\". Items are ## N. headings; a heading","icon":"$undefined","url":"/docs/usage/run-recipes/backlog-tasks--whats-pending-beyond-run-memory","$ref":"usage/run-recipes/backlog-tasks--whats-pending-beyond-run-memory.md"},{"$id":"usage/run-recipes/claude-code-memory-adapter-cyberine_cc_memory--opt-in.md","type":"page","name":"Claude Code memory adapter (CYBERINE_CC_MEMORY) — opt-in","description":"Cyberine's memory is its own: ~/.cyberine/memory/ (the Long-term memory section above), one format, vendor-independent, so the same orchestrator serves Claude, Codex or any other CLI. A Claude...","icon":"$undefined","url":"/docs/usage/run-recipes/claude-code-memory-adapter-cyberine_cc_memory--opt-in","$ref":"usage/run-recipes/claude-code-memory-adapter-cyberine_cc_memory--opt-in.md"},{"$id":"usage/run-recipes/shared-memory-backend--contextq-or-any-similar-server.md","type":"page","name":"Shared memory backend — ContextQ (or any similar server)","description":"The local episodic store above is the canonical record store and is always on. An OPTIONAL augmentation backend, selected purely by config, ALSO receives each terminal run and serves a...","icon":"$undefined","url":"/docs/usage/run-recipes/shared-memory-backend--contextq-or-any-similar-server","$ref":"usage/run-recipes/shared-memory-backend--contextq-or-any-similar-server.md"},{"$id":"usage/run-recipes/contextbrain--code-context-3-act-adapter-localbrain--contextqbrain-t441t454.md","type":"page","name":"ContextBrain — Code-Context 3-Act adapter (LocalBrain / ContextQBrain, T441/T454)","description":"Separate from the memory-preamble backend above, the ContextBrain port (src/core/contextBrain.ts) is the seam behind the Code-Context 3-Act architecture (docs/design/code-context-3act.md) used by","icon":"$undefined","url":"/docs/usage/run-recipes/contextbrain--code-context-3-act-adapter-localbrain--contextqbrain-t441t454","$ref":"usage/run-recipes/contextbrain--code-context-3-act-adapter-localbrain--contextqbrain-t441t454.md"},{"$id":"usage/run-recipes/project-groups--link-dirs-to-share-runs--chats--memory.md","type":"page","name":"Project groups — link dirs to share runs / chats / memory","description":"A GROUP links several project directories so their runs, chats, and long-term memory collaborate across projects (one Cyberine group = one shared memory pool). Groups are OPT-IN: with no config,...","icon":"$undefined","url":"/docs/usage/run-recipes/project-groups--link-dirs-to-share-runs--chats--memory","$ref":"usage/run-recipes/project-groups--link-dirs-to-share-runs--chats--memory.md"},{"$id":"usage/run-recipes/knowledge-ledger--cyberine-knowledge-k6tbw.md","type":"page","name":"Knowledge ledger — cyberine knowledge (k6tbw)","description":"An append-only ledger of lessons, facts, pointers and prefs that lives outside every repo, under $CYBERINEHOME/knowledge/ (default ~/.cyberine/knowledge/). Nothing is written inside the repo...","icon":"$undefined","url":"/docs/usage/run-recipes/knowledge-ledger--cyberine-knowledge-k6tbw","$ref":"usage/run-recipes/knowledge-ledger--cyberine-knowledge-k6tbw.md"},{"$id":"usage/run-recipes/expose-cyberine-as-an-mcp-server--npm-run-mcp.md","type":"page","name":"Expose Cyberine as an MCP server — npm run mcp","description":"npm run mcp starts Cyberine as an MCP server (stdio transport) so any MCP host (Claude Code, Cursor, Claude Desktop) can drive it. The server declares a short instructions string (a few lines: what","icon":"$undefined","url":"/docs/usage/run-recipes/expose-cyberine-as-an-mcp-server--npm-run-mcp","$ref":"usage/run-recipes/expose-cyberine-as-an-mcp-server--npm-run-mcp.md"},{"$id":"usage/run-recipes/consume-external-mcp-servers-as-tools--mcptooladapter.md","type":"page","name":"Consume external MCP servers as tools — McpToolAdapter","description":"Cyberine can consume any external MCP server's tools as orchestrator Tools (the consume-side mirror of npm run mcp, which exposes Cyberine itself). Configure servers by env; nothing connects when...","icon":"$undefined","url":"/docs/usage/run-recipes/consume-external-mcp-servers-as-tools--mcptooladapter","$ref":"usage/run-recipes/consume-external-mcp-servers-as-tools--mcptooladapter.md"},{"$id":"usage/run-recipes/oversized-tool-results--spill-to-disk--retrieval-spill_retrieve.md","type":"page","name":"Oversized tool results — spill-to-disk + retrieval (spill_retrieve)","description":"Every tool call an agent adapter makes (LoopAgentAdapter, AnthropicAgentAdapter, BedrockAdapter, VertexAdapter — including MCP tools consumed via McpToolAdapter, which previously had no cap at...","icon":"$undefined","url":"/docs/usage/run-recipes/oversized-tool-results--spill-to-disk--retrieval-spill_retrieve","$ref":"usage/run-recipes/oversized-tool-results--spill-to-disk--retrieval-spill_retrieve.md"},{"$id":"usage/run-recipes/history-compaction-spills-oversized-tool-results-instead-of-truncating-to-80-chars.md","type":"page","name":"History compaction spills oversized tool results instead of truncating to 80 chars","description":"For a worker with historyCompression: true on its WorkerProfile (see src/core/workerProfiles.ts), once accumulated message history exceeds historyCompressionBudget (default 8000 tokens,...","icon":"$undefined","url":"/docs/usage/run-recipes/history-compaction-spills-oversized-tool-results-instead-of-truncating-to-80-chars","$ref":"usage/run-recipes/history-compaction-spills-oversized-tool-results-instead-of-truncating-to-80-chars.md"},{"$id":"usage/run-recipes/malformed-tool-calls--validate-then-repair-on-the-loop-tool-input-path-kmk7x.md","type":"page","name":"Malformed tool calls — validate-then-repair on the loop tool-input path (kmk7x)","description":"Open models on OpenAI-compatible endpoints (DeepSeek, MiniMax, GLM, Qwen) emit a small set of recurring tool-call shape errors. Before this change a loop worker parsed the arguments with...","icon":"$undefined","url":"/docs/usage/run-recipes/malformed-tool-calls--validate-then-repair-on-the-loop-tool-input-path-kmk7x","$ref":"usage/run-recipes/malformed-tool-calls--validate-then-repair-on-the-loop-tool-input-path-kmk7x.md"},{"$id":"usage/run-recipes/edit-pacing--a-loop-implementer-writes-before-its-round-budget-runs-out.md","type":"page","name":"Edit pacing — a loop implementer writes before its round budget runs out","description":"A key-only loop implementer (loop-deepseek-build and the other loop--build workers) used to spend every tool round reading and never call a write tool. When the budget ran out, the loop asked for...","icon":"$undefined","url":"/docs/usage/run-recipes/edit-pacing--a-loop-implementer-writes-before-its-round-budget-runs-out","$ref":"usage/run-recipes/edit-pacing--a-loop-implementer-writes-before-its-round-budget-runs-out.md"},{"$id":"usage/run-recipes/repeat-tool-nudge--advisory-reminder-when-a-worker-repeats-the-exact-same-call.md","type":"page","name":"Repeat-tool nudge — advisory reminder when a worker repeats the exact same call","description":"safetyRails.checkPreCallCaps (iteration + token caps) runs once PER DISPATCH; a worker grepping the same pattern forty times inside ONE dispatch is invisible to it, and the only thing that...","icon":"$undefined","url":"/docs/usage/run-recipes/repeat-tool-nudge--advisory-reminder-when-a-worker-repeats-the-exact-same-call","$ref":"usage/run-recipes/repeat-tool-nudge--advisory-reminder-when-a-worker-repeats-the-exact-same-call.md"},{"$id":"usage/run-recipes/durable-worker-transcripts--cyberine-transcript.md","type":"page","name":"Durable worker transcripts — cyberine transcript","description":"A loop worker's events.jsonl is telemetry, not a transcript: each entry's detail is a 120-character slice of the prompt, so when a fleet worker fails there was previously no way to see what it was...","icon":"$undefined","url":"/docs/usage/run-recipes/durable-worker-transcripts--cyberine-transcript","$ref":"usage/run-recipes/durable-worker-transcripts--cyberine-transcript.md"},{"$id":"usage/run-recipes/capability-aware-routing--cyberine-route.md","type":"page","name":"Capability-aware routing — cyberine route","description":"Cyberine routes each task by marginal cost so it works standalone (a Claude Code / Codex replacement) on whatever you already pay for. Every worker carries a costModel and a priorityRank...","icon":"$undefined","url":"/docs/usage/run-recipes/capability-aware-routing--cyberine-route","$ref":"usage/run-recipes/capability-aware-routing--cyberine-route.md"},{"$id":"usage/run-recipes/smart-llm-as-router-for-low-confidence-tasks-t61.md","type":"page","name":"Smart (LLM-as-router) for low-confidence tasks (T61)","description":"When the deterministic router is unsure about a task (confidence < 0.7 — the \"default fallback\" bucket), you can opt in to a secondary LLM call that refines the worker ordering and per-worker...","icon":"$undefined","url":"/docs/usage/run-recipes/smart-llm-as-router-for-low-confidence-tasks-t61","$ref":"usage/run-recipes/smart-llm-as-router-for-low-confidence-tasks-t61.md"},{"$id":"usage/run-recipes/router-eval--compare-routing-strategies-on-the-corpus-t66.md","type":"page","name":"Router eval — compare routing strategies on the corpus (T66)","description":"Replays .local/runs/ offline and compares three routing strategies against each step's recorded judge verdict:","icon":"$undefined","url":"/docs/usage/run-recipes/router-eval--compare-routing-strategies-on-the-corpus-t66","$ref":"usage/run-recipes/router-eval--compare-routing-strategies-on-the-corpus-t66.md"},{"$id":"usage/run-recipes/model-strength-table-and-fleetjson-override-kyhae.md","type":"page","name":"Model strength table and fleet.json override (kyhae)","description":"Cyberine ships a sourced answer to \"which model is strong and which is weak\": every model id the fleet can run has a coarse rank from 1 to 5 (5 = strongest) and a cost class from 1 to 3, keyed by...","icon":"$undefined","url":"/docs/usage/run-recipes/model-strength-table-and-fleetjson-override-kyhae","$ref":"usage/run-recipes/model-strength-table-and-fleetjson-override-kyhae.md"},{"$id":"usage/run-recipes/fleet-bench--solve-rate-and-dollars-per-task-kvxah.md","type":"page","name":"Fleet bench — solve rate and dollars per task (kvxah)","description":"cyberine eval bench scores every available worker, and the judged winner, on a versioned suite of coding tasks with programmatic pass/fail. Router eval above compares routing decisions against...","icon":"$undefined","url":"/docs/usage/run-recipes/fleet-bench--solve-rate-and-dollars-per-task-kvxah","$ref":"usage/run-recipes/fleet-bench--solve-rate-and-dollars-per-task-kvxah.md"},{"$id":"usage/run-recipes/history-replay-bench-k0t77.md","type":"page","name":"History replay bench (k0t77)","description":"cyberine eval replay discovers past repository commits that added a scripts/smoke.ts grader and changed src/, starting on 2026-08-01. Each task starts from a git archive export of the parent...","icon":"$undefined","url":"/docs/usage/run-recipes/history-replay-bench-k0t77","$ref":"usage/run-recipes/history-replay-bench-k0t77.md"},{"$id":"usage/run-recipes/chat-prompt-cache-bench-kkrsb.md","type":"page","name":"Chat prompt-cache bench (kkrsb)","description":"npm run bench:chat-cache replays scripted chat turns through the real plannerTurn loop (real system prompt, real tool list, real read-only tools) and reads the provider's own cached-token count...","icon":"$undefined","url":"/docs/usage/run-recipes/chat-prompt-cache-bench-kkrsb","$ref":"usage/run-recipes/chat-prompt-cache-bench-kkrsb.md"},{"$id":"usage/run-recipes/llm-candidate-re-ranker--act-1-scope-filter-t446.md","type":"page","name":"LLM candidate re-ranker — Act 1 SCOPE filter (T446)","description":"An optional one-call sonnet-tier filter that runs during LocalBrain.boot() (Act 1 SCOPE). After the task-text extractor and git co-change ranker assemble a combined ~15-file candidate set, the...","icon":"$undefined","url":"/docs/usage/run-recipes/llm-candidate-re-ranker--act-1-scope-filter-t446","$ref":"usage/run-recipes/llm-candidate-re-ranker--act-1-scope-filter-t446.md"},{"$id":"usage/run-recipes/per-task-model-variant-selection-t60.md","type":"page","name":"Per-task model variant selection (T60)","description":"Loop workers that expose a variants map in their WorkerProfile run a different model id depending on task tier — the router picks the cheapest capable variant so you don't pay for deep reasoning...","icon":"$undefined","url":"/docs/usage/run-recipes/per-task-model-variant-selection-t60","$ref":"usage/run-recipes/per-task-model-variant-selection-t60.md"},{"$id":"usage/run-recipes/machine-readable-route-recommendation--cyberine-route---json-t62.md","type":"page","name":"Machine-readable route recommendation — cyberine route --json (T62)","description":"cyberine route --json \"<task>\" emits a stable JSON object designed for non-Cyberine callers (e.g. a Claude Code subagent dispatcher) that need to ask Cyberine \"what tier should this task run at\"...","icon":"$undefined","url":"/docs/usage/run-recipes/machine-readable-route-recommendation--cyberine-route---json-t62","$ref":"usage/run-recipes/machine-readable-route-recommendation--cyberine-route---json-t62.md"},{"$id":"usage/run-recipes/routing-explain--why-a-worker-was-picked-or-skipped-kc0nz.md","type":"page","name":"Routing-explain — why a worker was picked or skipped (kc0nz)","description":"Routing used to be a black box once the decision was made: reasons[] explained the winning fleet, but nothing recorded which OTHER candidates existed, or why a worker never even reached the router...","icon":"$undefined","url":"/docs/usage/run-recipes/routing-explain--why-a-worker-was-picked-or-skipped-kc0nz","$ref":"usage/run-recipes/routing-explain--why-a-worker-was-picked-or-skipped-kc0nz.md"},{"$id":"usage/run-recipes/opentelemetry-genai-export--cyberine_otel_endpoint-kyv0t.md","type":"page","name":"OpenTelemetry GenAI export — CYBERINE_OTEL_ENDPOINT (kyv0t)","description":"run.json/events.jsonl is a proprietary, Cyberine-only format with no interop. Setting CYBERINEOTELENDPOINT turns on an exporter that maps RunStore run/step/worker events onto spans and ships them...","icon":"$undefined","url":"/docs/usage/run-recipes/opentelemetry-genai-export--cyberine_otel_endpoint-kyv0t","$ref":"usage/run-recipes/opentelemetry-genai-export--cyberine_otel_endpoint-kyv0t.md"},{"$id":"usage/run-recipes/swarm-mode--cyberine-run---mode-swarm-plan---parallel-chunk-implement---merge.md","type":"page","name":"Swarm mode — cyberine run --mode swarm (plan -> parallel-chunk implement -> merge)","description":"For tasks that decompose cleanly into disjoint chunks — large multi-file refactors, sweep updates across many files, any scope where one worker would burn tool rounds but N workers can split the...","icon":"$undefined","url":"/docs/usage/run-recipes/swarm-mode--cyberine-run---mode-swarm-plan---parallel-chunk-implement---merge","$ref":"usage/run-recipes/swarm-mode--cyberine-run---mode-swarm-plan---parallel-chunk-implement---merge.md"},{"$id":"usage/run-recipes/lane-grouping--when-lanes-help-vs-flat-fan-out.md","type":"page","name":"Lane grouping — when lanes help vs flat fan-out","description":"The T269 lane runner is the default swarm executor in --mode swarm. It runs the plan through clusterChunks + packLanes BEFORE workers fan out, so the chunks a worker can see on disk follow the...","icon":"$undefined","url":"/docs/usage/run-recipes/lane-grouping--when-lanes-help-vs-flat-fan-out","$ref":"usage/run-recipes/lane-grouping--when-lanes-help-vs-flat-fan-out.md"},{"$id":"usage/run-recipes/operating-modes----mode-competeswarmdeepsmartrushharvest--cyberine_mode-env.md","type":"page","name":"Operating modes — --mode compete|swarm|deep|smart|rush|harvest (+ CYBERINE_MODE env)","description":"Six preset bundles (PRESETS in src/core/modePresets.ts) that drive WorkflowEngine behaviour end-to-end: fleet size cap, judge chain selection, verify gate, governor, and M5 escalation. Selected...","icon":"$undefined","url":"/docs/usage/run-recipes/operating-modes----mode-competeswarmdeepsmartrushharvest--cyberine_mode-env","$ref":"usage/run-recipes/operating-modes----mode-competeswarmdeepsmartrushharvest--cyberine_mode-env.md"},{"$id":"usage/run-recipes/research-mode--cyberine-research-plan---search---verify---synthesize.md","type":"page","name":"Research mode — cyberine research (plan -> search -> verify -> synthesize)","description":"For questions that need current web evidence — comparisons, market scans, \"what is the state of X in 2026?\", trade-off write-ups — use cyberine research. The orchestrator decomposes the question...","icon":"$undefined","url":"/docs/usage/run-recipes/research-mode--cyberine-research-plan---search---verify---synthesize","$ref":"usage/run-recipes/research-mode--cyberine-research-plan---search---verify---synthesize.md"},{"$id":"usage/run-recipes/prior-art-research--cyberine-research---prior-art-ks0sh.md","type":"page","name":"Prior-art research — cyberine research --prior-art (ks0sh)","description":"Competitor and alternatives research that ends in a feature parity matrix and a recommendations ledger, where every recommendation leaves the report as a task id or a dated rejection reason.","icon":"$undefined","url":"/docs/usage/run-recipes/prior-art-research--cyberine-research---prior-art-ks0sh","$ref":"usage/run-recipes/prior-art-research--cyberine-research---prior-art-ks0sh.md"},{"$id":"usage/run-recipes/website-rebuild-target-nextjs--fumadocs--site.md","type":"page","name":"Website rebuild target (Next.js + Fumadocs) — site/","description":"The rebuild (T666) lives in site/ as a separate buildable target, and since the T670 cutover (2026-09-25) it is the only docs site.","icon":"$undefined","url":"/docs/usage/run-recipes/website-rebuild-target-nextjs--fumadocs--site","$ref":"usage/run-recipes/website-rebuild-target-nextjs--fumadocs--site.md"},{"$id":"usage/run-recipes/documentation-site-vitepress-retired-t670.md","type":"page","name":"Documentation site: VitePress retired (T670)","description":"The VitePress implementation (docs/.vitepress, npm run docs:legacy:) was deleted on 2026-09-25 after the new site passed its full QA gate: the static site:qa pass (0 HIGH, 0 dead links across 59...","icon":"$undefined","url":"/docs/usage/run-recipes/documentation-site-vitepress-retired-t670","$ref":"usage/run-recipes/documentation-site-vitepress-retired-t670.md"},{"$id":"usage/run-recipes/fleet-policy-file--cyberine-fleet-t422-t425-t426.md","type":"page","name":"Fleet policy file — cyberine fleet (T422, T425, T426)","description":"fleet.json is an optional provider policy file that declares which workers are active and which model handles each task kind. It externalizes the routing that is otherwise hardcoded across worker...","icon":"$undefined","url":"/docs/usage/run-recipes/fleet-policy-file--cyberine-fleet-t422-t425-t426","$ref":"usage/run-recipes/fleet-policy-file--cyberine-fleet-t422-t425-t426.md"},{"$id":"usage/run-recipes/unified-health-check--cyberine-doctor-t540.md","type":"page","name":"Unified health check — cyberine doctor (T540)","description":"One report covering the whole install, printed as pass/warn/fail lines with a fix hint on every warn/fail:","icon":"$undefined","url":"/docs/usage/run-recipes/unified-health-check--cyberine-doctor-t540","$ref":"usage/run-recipes/unified-health-check--cyberine-doctor-t540.md"},{"$id":"usage/run-recipes/adapter-liveness--is-it-funded-not-just-keyed---liveness-kwrp9.md","type":"page","name":"Adapter liveness — is it FUNDED, not just keyed (--liveness, kwrp9)","description":"cyberine doctor answers \"is a key present\". It never answered \"is that key's account funded\", and neither did the fleet TUI's overview, which listed an adapter whenever its key existed and printed...","icon":"$undefined","url":"/docs/usage/run-recipes/adapter-liveness--is-it-funded-not-just-keyed---liveness-kwrp9","$ref":"usage/run-recipes/adapter-liveness--is-it-funded-not-just-keyed---liveness-kwrp9.md"},{"$id":"usage/run-recipes/selftest-sweep-with-a-real-skip-channel--npm-run-sweep-t707.md","type":"page","name":"Selftest sweep with a real SKIP channel — npm run sweep (T707)","description":"This project has no test framework; the selftest: smokes ARE the test suite, and the sweep is how you run all of them, on the operator's word, typically before a promotion. It replaced ~15 lines...","icon":"$undefined","url":"/docs/usage/run-recipes/selftest-sweep-with-a-real-skip-channel--npm-run-sweep-t707","$ref":"usage/run-recipes/selftest-sweep-with-a-real-skip-channel--npm-run-sweep-t707.md"},{"$id":"usage/run-recipes/security-gate-drift-check--npm-run-selftestgate-list-drift-t707-retargeted-2026-09-07.md","type":"page","name":"Security-gate drift check — npm run selftest:gate-list-drift (T707, retargeted 2026-09-07)","description":"The security/data-integrity smoke list froze on 2026-08-22 in two hand-maintained copies (the CI gate loop and the git-workflow rule's precondition 3), and ten security smokes shipped afterwards...","icon":"$undefined","url":"/docs/usage/run-recipes/security-gate-drift-check--npm-run-selftestgate-list-drift-t707-retargeted-2026-09-07","$ref":"usage/run-recipes/security-gate-drift-check--npm-run-selftestgate-list-drift-t707-retargeted-2026-09-07.md"},{"$id":"usage/run-recipes/promotion-check--npm-run-promotecheck-2026-10-02.md","type":"page","name":"Promotion check — npm run promote:check (2026-10-02)","description":"Operator decision 2026-10-02: no customers yet and still in the dev phase, so there is no automatic CI. .github/workflows/ci.yml and .githooks/pre-push were deleted, the homelab self-hosted runner...","icon":"$undefined","url":"/docs/usage/run-recipes/promotion-check--npm-run-promotecheck-2026-10-02","$ref":"usage/run-recipes/promotion-check--npm-run-promotecheck-2026-10-02.md"},{"$id":"usage/run-recipes/sweep-stamp--localselftest-stamp-2026-09-07-push-gate-removed-2026-09-30-hook-removed-2026-10-02.md","type":"page","name":"Sweep stamp — .local/selftest-stamp/ (2026-09-07; push gate removed 2026-09-30, hook removed 2026-10-02)","description":"A full npm run sweep (no --only / --security) records its verdict in .local/selftest-stamp/<host>/<HEAD sha>.json. From 2026-09-07 a tracked pre-push hook refused a push to main without an ok:...","icon":"$undefined","url":"/docs/usage/run-recipes/sweep-stamp--localselftest-stamp-2026-09-07-push-gate-removed-2026-09-30-hook-removed-2026-10-02","$ref":"usage/run-recipes/sweep-stamp--localselftest-stamp-2026-09-07-push-gate-removed-2026-09-30-hook-removed-2026-10-02.md"},{"$id":"usage/run-recipes/sweep-flake-ledger--npm-run-sweepflakes-kdpxk-2026-09-28.md","type":"page","name":"Sweep flake ledger — npm run sweep:flakes (kdpxk, 2026-09-28)","description":"Every sweep, full or partial, appends one line to .local/selftest-stamp/<host>/ledger.jsonl: at, sha, host, scope (full / security / only, with selected for a partial run), the pass count,...","icon":"$undefined","url":"/docs/usage/run-recipes/sweep-flake-ledger--npm-run-sweepflakes-kdpxk-2026-09-28","$ref":"usage/run-recipes/sweep-flake-ledger--npm-run-sweepflakes-kdpxk-2026-09-28.md"},{"$id":"usage/run-recipes/dependency-audit-gate--npm-run-selftestdep-audit-t673-multi-tree-since-t707.md","type":"page","name":"Dependency audit gate — npm run selftest:dep-audit (T673, multi-tree since T707)","description":"Recurring guard against silently-accruing vulnerable dependencies (the gap that let fast-uri / ip-address / nanoid / postcss sit at HIGH for weeks before 2026-08-20). It discovers every...","icon":"$undefined","url":"/docs/usage/run-recipes/dependency-audit-gate--npm-run-selftestdep-audit-t673-multi-tree-since-t707","$ref":"usage/run-recipes/dependency-audit-gate--npm-run-selftestdep-audit-t673-multi-tree-since-t707.md"},{"$id":"usage/run-recipes/credit-ledger--track-provider-balances-gate-low-credit-workers.md","type":"page","name":"Credit ledger — track provider balances, gate low-credit workers","description":"Cyberine can track an approximate estimated remaining balance per provider and skip or down-rank workers whose provider is running low. Credits do not need to match the provider dashboard exactly;...","icon":"$undefined","url":"/docs/usage/run-recipes/credit-ledger--track-provider-balances-gate-low-credit-workers","$ref":"usage/run-recipes/credit-ledger--track-provider-balances-gate-low-credit-workers.md"},{"$id":"usage/run-recipes/tui-stdout-shim--synchronized-output-and-how-to-smoke-it.md","type":"page","name":"TUI stdout shim — synchronized output, and how to smoke it","description":"Ink repaints the whole frame every tick. makeSyncStdout() (src/tui/syncOutput.ts) proxies the render stream and wraps each write in the DEC 2026 begin/end synchronized-update markers, so a...","icon":"$undefined","url":"/docs/usage/run-recipes/tui-stdout-shim--synchronized-output-and-how-to-smoke-it","$ref":"usage/run-recipes/tui-stdout-shim--synchronized-output-and-how-to-smoke-it.md"},{"$id":"usage/run-recipes/task-footprint-estimator--history-driven-budget-planning-kx9rt-calibrated-in-k5m89.md","type":"page","name":"Task footprint estimator — history-driven budget planning (kx9rt, calibrated in k5m89)","description":"planBudget() decides whether a step can spawn all its workers, must chunk into a handoff chain, should degrade, or has to abort. Its cost side comes from the credit ledger above; its SIZE side...","icon":"$undefined","url":"/docs/usage/run-recipes/task-footprint-estimator--history-driven-budget-planning-kx9rt-calibrated-in-k5m89","$ref":"usage/run-recipes/task-footprint-estimator--history-driven-budget-planning-kx9rt-calibrated-in-k5m89.md"},{"$id":"usage/run-recipes/usage-records-for-external-meters--billed-dollars-per-model-tokens-aux-spend-k0nnx-kq39p-kcajh.md","type":"page","name":"Usage records for external meters — billed dollars, per-model tokens, aux spend (k0nnx, kq39p, kcajh)","description":"Every dollar figure Cyberine keeps now travels with the model and the token counts behind it, so a meter outside Cyberine (tokenburn, a spreadsheet) can price it again or check it without...","icon":"$undefined","url":"/docs/usage/run-recipes/usage-records-for-external-meters--billed-dollars-per-model-tokens-aux-spend-k0nnx-kq39p-kcajh","$ref":"usage/run-recipes/usage-records-for-external-meters--billed-dollars-per-model-tokens-aux-spend-k0nnx-kq39p-kcajh.md"},{"$id":"usage/run-recipes/hard-spend-cap--opt-in-race-free-ceiling-across-all-providers.md","type":"page","name":"Hard spend cap — opt-in, race-free ceiling across ALL providers","description":"CYBERINECREDITMINUSD above is a SOFT, per-provider gate: it reads the folded balance and down-ranks/skips a worker, but two concurrent dispatches can each read \"balance ok\" moments apart and,...","icon":"$undefined","url":"/docs/usage/run-recipes/hard-spend-cap--opt-in-race-free-ceiling-across-all-providers","$ref":"usage/run-recipes/hard-spend-cap--opt-in-race-free-ceiling-across-all-providers.md"},{"$id":"usage/run-recipes/quota-cooldown--automatic-circuit-breaker-after-402429.md","type":"page","name":"Quota cooldown — automatic circuit-breaker after 402/429","description":"When a worker fails with a quota signal (HTTP 402 payment required / insufficient credits, or HTTP 429 rate-limit / too-many-requests), Cyberine records a temporary cooldown for that worker ID in...","icon":"$undefined","url":"/docs/usage/run-recipes/quota-cooldown--automatic-circuit-breaker-after-402429","$ref":"usage/run-recipes/quota-cooldown--automatic-circuit-breaker-after-402429.md"},{"$id":"usage/run-recipes/blocker-visibility--waitingfor-on-stalled-workers.md","type":"page","name":"Blocker visibility — waitingFor on stalled workers","description":"When a worker cannot make progress, a waitingFor field explains why. It appears in:","icon":"$undefined","url":"/docs/usage/run-recipes/blocker-visibility--waitingfor-on-stalled-workers","$ref":"usage/run-recipes/blocker-visibility--waitingfor-on-stalled-workers.md"},{"$id":"usage/run-recipes/provider-circuit-breaker-and-bounded-retry--one-dead-provider-degrades-the-fleet-not-the-run.md","type":"page","name":"Provider circuit breaker and bounded retry — one dead provider degrades the fleet, not the run","description":"Distinct from the quota cooldown above, which reacts only to a 402/429 quota signal: the breaker counts consecutive terminal failures of any shape at the Orchestrator.dispatch() seam, so it also...","icon":"$undefined","url":"/docs/usage/run-recipes/provider-circuit-breaker-and-bounded-retry--one-dead-provider-degrades-the-fleet-not-the-run","$ref":"usage/run-recipes/provider-circuit-breaker-and-bounded-retry--one-dead-provider-degrades-the-fleet-not-the-run.md"},{"$id":"usage/run-recipes/fallback-model-chain--per-worker-model-degradation.md","type":"page","name":"Fallback model chain — per-worker model degradation","description":"When a LoopAgentAdapter worker's primary model is overloaded or unavailable (HTTP 429 / 402 / 5xx / unexpected overload 400), Cyberine retries the same request with the next model in a configured...","icon":"$undefined","url":"/docs/usage/run-recipes/fallback-model-chain--per-worker-model-degradation","$ref":"usage/run-recipes/fallback-model-chain--per-worker-model-degradation.md"},{"$id":"usage/run-recipes/cli-flag-compat-diagnosis--structured-output-knobs-t656.md","type":"page","name":"CLI flag-compat diagnosis + structured-output knobs (T656)","description":"The claude and codex build workers pass version-gated CLI flags (e.g. codex's --json, codex's --dangerously-bypass-hook-trust when CYBERINECODEXBYPASSHOOKTRUST=1 opts in, claude's...","icon":"$undefined","url":"/docs/usage/run-recipes/cli-flag-compat-diagnosis--structured-output-knobs-t656","$ref":"usage/run-recipes/cli-flag-compat-diagnosis--structured-output-knobs-t656.md"},{"$id":"usage/run-recipes/claude-cold-spawn-session-persistence-suppression-t627.md","type":"page","name":"Claude cold-spawn session-persistence suppression (T627)","description":"Every cold-spawn claude build-worker dispatch left a resumable session transcript on disk under the operator's own ~/.claude session store, even though Cyberine never resumes a cold-spawn worker —...","icon":"$undefined","url":"/docs/usage/run-recipes/claude-cold-spawn-session-persistence-suppression-t627","$ref":"usage/run-recipes/claude-cold-spawn-session-persistence-suppression-t627.md"},{"$id":"usage/run-recipes/claude-prompt-answering----permission-prompts-none-k7hha.md","type":"page","name":"Claude prompt answering — --permission-prompts none (k7hha)","description":"A claude build worker is spawned with stdin CLOSED and with neither an SDK host nor --permission-prompt-tool. Claude Code's --permission-prompts flag (v2.1.259) defaults to host, which routes any...","icon":"$undefined","url":"/docs/usage/run-recipes/claude-prompt-answering----permission-prompts-none-k7hha","$ref":"usage/run-recipes/claude-prompt-answering----permission-prompts-none-k7hha.md"},{"$id":"usage/run-recipes/a-non-success-result-subtype-is-a-failed-run-kq5rd.md","type":"page","name":"A non-success result subtype is a failed run (kq5rd)","description":"parseClaudeStreamJson used to take the last type:'result' event and read its result field without looking at subtype. An error subtype carries no result (Claude Code puts errors[] there instead),...","icon":"$undefined","url":"/docs/usage/run-recipes/a-non-success-result-subtype-is-a-failed-run-kq5rd","$ref":"usage/run-recipes/a-non-success-result-subtype-is-a-failed-run-kq5rd.md"},{"$id":"usage/run-recipes/where-a-workers-claude_code_-settings-came-from--toolingsnapshotworkerenv-kg4pt.md","type":"page","name":"Where a worker's CLAUDE_CODE_* settings came from — toolingSnapshot.workerEnv (kg4pt)","description":"A spawned claude worker inherits the orchestrator's environment verbatim apart from provider credentials (buildScopedChildEnv). A Claude Code session exports its settings.json env block, so a...","icon":"$undefined","url":"/docs/usage/run-recipes/where-a-workers-claude_code_-settings-came-from--toolingsnapshotworkerenv-kg4pt","$ref":"usage/run-recipes/where-a-workers-claude_code_-settings-came-from--toolingsnapshotworkerenv-kg4pt.md"},{"$id":"usage/run-recipes/validate-a-harness-overlay--cyberine-harness-check-kdgkp.md","type":"page","name":"Validate a harness overlay — cyberine harness check (kdgkp)","description":"A harness overlay is a directory of proposed edits to what Cyberine injects into workers: rules/<name>.md (replaces or adds a project rule file in .cyberine/rules/), skills/<id>/SKILL.md and files...","icon":"$undefined","url":"/docs/usage/run-recipes/validate-a-harness-overlay--cyberine-harness-check-kdgkp","$ref":"usage/run-recipes/validate-a-harness-overlay--cyberine-harness-check-kdgkp.md"},{"$id":"usage/run-recipes/one-harness-repair-after-a-verify-failure--cyberine_harness_repair-k5dc7.md","type":"page","name":"One harness repair after a verify failure — CYBERINE_HARNESS_REPAIR (k5dc7)","description":"Off by default (operator decision D3-A). With CYBERINEHARNESSREPAIR=1, a code task that cyberine autopilot runs on its fleet engine with sync dispatch (the default; --dispatch sync or --verify)...","icon":"$undefined","url":"/docs/usage/run-recipes/one-harness-repair-after-a-verify-failure--cyberine_harness_repair-k5dc7","$ref":"usage/run-recipes/one-harness-repair-after-a-verify-failure--cyberine_harness_repair-k5dc7.md"},{"$id":"usage/run-recipes/contrastive-reflection-into-a-lesson-library--cyberine-lessons-k79pw.md","type":"page","name":"Contrastive reflection into a lesson library — cyberine lessons (k79pw)","description":"Reflect on one completed step only when the record contains a passed worker and a failed worker with the same brief. A judge selecting one of several successful workers is not a failure signal....","icon":"$undefined","url":"/docs/usage/run-recipes/contrastive-reflection-into-a-lesson-library--cyberine-lessons-k79pw","$ref":"usage/run-recipes/contrastive-reflection-into-a-lesson-library--cyberine-lessons-k79pw.md"},{"$id":"usage/run-recipes/which-harness-a-worker-ran-with--toolingsnapshotharness-k7xxg.md","type":"page","name":"Which harness a worker ran with — toolingSnapshot.harness (k7xxg)","description":"Every engine run records, in run.json, a hash of each harness input Cyberine injected, so two runs can be compared on the harness they used without storing the text itself (instruction files can...","icon":"$undefined","url":"/docs/usage/run-recipes/which-harness-a-worker-ran-with--toolingsnapshotharness-k7xxg","$ref":"usage/run-recipes/which-harness-a-worker-ran-with--toolingsnapshotharness-k7xxg.md"},{"$id":"usage/run-recipes/which-claude-account-a-worker-bills--oauthtoken-k8rhz.md","type":"page","name":"Which Claude account a worker bills — oauthToken (k8rhz)","description":"CLAUDECODEOAUTHTOKEN is a long-lived subscription credential that decides WHICH SEAT a claude worker bills. It used to reach every spawned worker by plain inheritance: T603's least-privilege env","icon":"$undefined","url":"/docs/usage/run-recipes/which-claude-account-a-worker-bills--oauthtoken-k8rhz","$ref":"usage/run-recipes/which-claude-account-a-worker-bills--oauthtoken-k8rhz.md"},{"$id":"usage/run-recipes/which-secrets-a-cli-worker-sees--generic-secret-strip-and-passthrough-kznbp.md","type":"page","name":"Which secrets a CLI worker sees — generic secret strip and passthrough (kznbp)","description":"T603 and k8rhz scoped provider credentials only, so every other secret in the launching shell reached an auto-approve CLI worker: GITHUBTOKEN, NPMTOKEN, AWSSECRETACCESSKEY, DATABASEURL,...","icon":"$undefined","url":"/docs/usage/run-recipes/which-secrets-a-cli-worker-sees--generic-secret-strip-and-passthrough-kznbp","$ref":"usage/run-recipes/which-secrets-a-cli-worker-sees--generic-secret-strip-and-passthrough-kznbp.md"},{"$id":"usage/run-recipes/claude-session-names--cy-action-run8-session8-seat.md","type":"page","name":"Claude session names — cy-<action>-<run8>-<session8>[-<seat>]","description":"Every claude session Cyberine spawns registers under a name that starts with cy-, so claude agents, ListAgents and ~/.claude/sessions/<pid>.json tell a Cyberine child apart from the operator's own...","icon":"$undefined","url":"/docs/usage/run-recipes/claude-session-names--cy-action-run8-session8-seat","$ref":"usage/run-recipes/claude-session-names--cy-action-run8-session8-seat.md"},{"$id":"usage/run-recipes/background-fleet-worker--claude---bg-k56eq.md","type":"page","name":"Background fleet worker — claude --bg (k56eq)","description":"The default claude build worker is an opaque claude -p child: nothing can open it, steer it, or outlive the orchestrator that spawned it. CYBERINECLAUDEWORKERMODE=bg swaps that ONE worker for a...","icon":"$undefined","url":"/docs/usage/run-recipes/background-fleet-worker--claude---bg-k56eq","$ref":"usage/run-recipes/background-fleet-worker--claude---bg-k56eq.md"},{"$id":"usage/run-recipes/idle-stream-watchdog--abort-a-stalled-stream-fleet-worker--chat-t562-t570.md","type":"page","name":"Idle-stream watchdog — abort a stalled stream, fleet worker + chat (T562, T570)","description":"Claude Code v2.1.196 made a streaming idle watchdog default-on: it aborts and retries a response stream that produces no events for 5 minutes. LoopAgentAdapter (fleet workers) AND the interactive...","icon":"$undefined","url":"/docs/usage/run-recipes/idle-stream-watchdog--abort-a-stalled-stream-fleet-worker--chat-t562-t570","$ref":"usage/run-recipes/idle-stream-watchdog--abort-a-stalled-stream-fleet-worker--chat-t562-t570.md"},{"$id":"usage/run-recipes/worker-internal-subagents--delegate_subtask-t313.md","type":"page","name":"Worker-internal subagents — delegate_subtask (T313)","description":"Implementer-class loop workers (loop--build) automatically get a delegatesubtask tool that lets the worker spawn a scoped subagent — a focused nested LoopAgentAdapter with its own system prompt, a...","icon":"$undefined","url":"/docs/usage/run-recipes/worker-internal-subagents--delegate_subtask-t313","$ref":"usage/run-recipes/worker-internal-subagents--delegate_subtask-t313.md"},{"$id":"usage/run-recipes/tool-call-pipeline--batch-multi-tool-sequences-in-one-model-turn-t319.md","type":"page","name":"Tool-call pipeline — batch multi-tool sequences in one model turn (T319)","description":"Implementer-class loop workers (loop--build) automatically get a runpipeline tool that lets a worker express a predictable multi-step tool sequence as a single tool call, collapsing N sequential...","icon":"$undefined","url":"/docs/usage/run-recipes/tool-call-pipeline--batch-multi-tool-sequences-in-one-model-turn-t319","$ref":"usage/run-recipes/tool-call-pipeline--batch-multi-tool-sequences-in-one-model-turn-t319.md"},{"$id":"usage/run-recipes/result-projection--configurable-worker-output-fidelity-t328.md","type":"page","name":"Result projection — configurable worker output fidelity (T328)","description":"By default every worker returns a full Result: the complete output text, usage stats, and any raw provider payload. When a judge or downstream step only needs a summary (e.g. an LLM judge that...","icon":"$undefined","url":"/docs/usage/run-recipes/result-projection--configurable-worker-output-fidelity-t328","$ref":"usage/run-recipes/result-projection--configurable-worker-output-fidelity-t328.md"},{"$id":"usage/run-recipes/stateful-session-reuse--warm-cli-agent-bindings--session-tree-ingestion-t333.md","type":"page","name":"Stateful session reuse — warm CLI-agent bindings + session tree ingestion (T333)","description":"By default every CLI worker cold-spawns a brand-new session for every task. With warm bindings enabled the adapter records the session ID returned by the first successful dispatch, then passes it...","icon":"$undefined","url":"/docs/usage/run-recipes/stateful-session-reuse--warm-cli-agent-bindings--session-tree-ingestion-t333","$ref":"usage/run-recipes/stateful-session-reuse--warm-cli-agent-bindings--session-tree-ingestion-t333.md"},{"$id":"usage/run-recipes/pluggable-execution-backends--run-workers-in-docker-or-on-a-remote-host-t332.md","type":"page","name":"Pluggable execution backends — run workers in Docker or on a remote host (T332)","description":"By default every isolate-step worker provisions a local git worktree and executes its CLI or loop work on the local host. The execution backend abstraction decouples WHERE the worker runs from...","icon":"$undefined","url":"/docs/usage/run-recipes/pluggable-execution-backends--run-workers-in-docker-or-on-a-remote-host-t332","$ref":"usage/run-recipes/pluggable-execution-backends--run-workers-in-docker-or-on-a-remote-host-t332.md"}],"$id":"usage/run-recipes","$ref":{"folder":"usage/run-recipes","meta":"usage/run-recipes/meta.json"},"icon":"$undefined"},{"$id":"usage/extend-recipes-the-vendor-independence-payoff.md","type":"page","name":"Extend recipes (the vendor-independence payoff)","description":"Three independent ways, none touch the core:","icon":"$undefined","url":"/docs/usage/extend-recipes-the-vendor-independence-payoff","$ref":"usage/extend-recipes-the-vendor-independence-payoff.md"},{"$id":"usage/durable-teammate.md","type":"page","name":"Durable teammate","description":"Enables two complementary approaches so a teammate can receive and act on messages even after it has gone idle or exited. Both share the same core handler (src/store/durableTeammate.ts). Default...","icon":"$undefined","url":"/docs/usage/durable-teammate","$ref":"usage/durable-teammate.md"},{"$id":"usage/cross-run-comms-over-http-hub-and-spoke-message-bus.md","type":"page","name":"Cross-run comms over HTTP (hub-and-spoke message bus)","description":"The default file bus backend is filesystem IPC: it reaches every process on ONE machine and nothing beyond it. CYBERINEBUSBACKEND=http swaps in a backend that carries the same publish / subscribe...","icon":"$undefined","url":"/docs/usage/cross-run-comms-over-http-hub-and-spoke-message-bus","$ref":"usage/cross-run-comms-over-http-hub-and-spoke-message-bus.md"},{"$id":"usage/background-shell-jobs-inside-a-worker-run_bash-background--bash_jobs.md","type":"page","name":"Background shell jobs inside a worker (run_bash background + bash_jobs)","description":"A loop worker can start a long-running shell command without blocking its tool round: runbash with background: true starts the command immediately and returns a job id; bashjobs with action=\"list\"...","icon":"$undefined","url":"/docs/usage/background-shell-jobs-inside-a-worker-run_bash-background--bash_jobs","$ref":"usage/background-shell-jobs-inside-a-worker-run_bash-background--bash_jobs.md"},{"$id":"usage/what-cyberine-writes-on-disk.md","type":"page","name":"What Cyberine writes on disk","description":"A reference of every path the product code creates, built from the writers in src/ (2026-10-18). \"Secret\" marks a file that can hold credentials or private text: keep it out of git, backups you...","icon":"$undefined","url":"/docs/usage/what-cyberine-writes-on-disk","$ref":"usage/what-cyberine-writes-on-disk.md"},{"$id":"usage/environment-variables.md","type":"page","name":"Environment variables","description":"Environment variables: a Cyberine usage recipe.","icon":"$undefined","url":"/docs/usage/environment-variables","$ref":"usage/environment-variables.md"},{"$id":"usage/generated-reference-do-not-edit-by-hand.md","type":"page","name":"Generated reference (do not edit by hand)","description":"Regenerate with npm run docs:catalog. npm run docs:catalog:check fails (non-zero exit, writes nothing) when source has drifted from these tables -- e.g. a new tool or env var was added and this...","icon":"$undefined","url":"/docs/usage/generated-reference-do-not-edit-by-hand","$ref":"usage/generated-reference-do-not-edit-by-hand.md"}],"$id":"usage","$ref":{"folder":"usage","meta":"usage/meta.json"},"icon":"$undefined"},{"$id":"tasks-manifest.md","type":"page","name":"Task manifest format","description":"This is the normative specification of the task-manifest format used by the cyberine orchestrator (src/tasks/manifestReader.ts, src/tasks/statusWriter.ts, src/tasks/runnable.ts on the read side;...","icon":"$undefined","url":"/docs/tasks-manifest","$ref":"tasks-manifest.md"},{"type":"folder","name":"Providers","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"providers/index.md","type":"page","name":"Provider API Reference","description":"A trace-it-fast reference for every AI provider / coding-CLI that Cyberine integrates directly. Each provider has its own page with the official API/CLI surface (endpoints, auth, models, gotchas,...","icon":"$undefined","url":"/docs/providers","$ref":"providers/index.md"},{"$id":"providers/changelog-watch.md","type":"page","name":"Upstream Changelog Watch List","description":"Every upstream Cyberine follows, in one table, so \"anything new upstream?\" resolves to the right changelog in one hop. URLs are pulled from each provider's page in this folder (its \"Official docs\"...","icon":"$undefined","url":"/docs/providers/changelog-watch","$ref":"providers/changelog-watch.md"},{"$id":"providers/model-strength.md","type":"page","name":"Model strength reference (2026-09-17)","description":"One sourced answer to \"which enabled model is strongest\", used by the router today and meant for planning, task splitting and judging. It ranks every model id the fleet can run (every...","icon":"$undefined","url":"/docs/providers/model-strength","$ref":"providers/model-strength.md"},{"$id":"providers/deepseek.md","type":"page","name":"DeepSeek API (direct)","description":"All paths are relative to https://api.deepseek.com. Every one below was called live on 2026-08-25 (T679) and returned the status shown.","icon":"$undefined","url":"/docs/providers/deepseek","$ref":"providers/deepseek.md"},{"$id":"providers/openrouter.md","type":"page","name":"OpenRouter API","description":"No per-provider auth — OpenRouter wraps every provider under one key.","icon":"$undefined","url":"/docs/providers/openrouter","$ref":"providers/openrouter.md"},{"$id":"providers/anthropic.md","type":"page","name":"Anthropic Claude API (Messages)","description":"Fable 5 $10/$50 · Opus 5.5 $4/$20 · Opus 5 + 4.x $5/$25 (Opus 5 / Opus 4.8 fast mode $10/$50) · Sonnet 5.5 and Sonnet 5 $2/$10 (permanent since 2026-08; the 2026-09-01 rise to $3/$15 was...","icon":"$undefined","url":"/docs/providers/anthropic","$ref":"providers/anthropic.md"},{"$id":"providers/openai.md","type":"page","name":"OpenAI API (Chat Completions + Responses)","description":"The openai Node SDK constructor accepts a custom baseURL — this is how Cyberine points the same client at LiteLLM/OpenRouter/DeepSeek without code changes.","icon":"$undefined","url":"/docs/providers/openai","$ref":"providers/openai.md"},{"$id":"providers/litellm.md","type":"page","name":"LiteLLM Proxy / Gateway","description":"These two defaults are placeholders that signal \"unconfigured\" — override them with real upstream ids. Underlying provider docs: openai.md, anthropic.md.","icon":"$undefined","url":"/docs/providers/litellm","$ref":"providers/litellm.md"},{"$id":"providers/moonshot-kimi.md","type":"page","name":"Moonshot AI / Kimi API","description":"A vendor-independent multi-provider AI agent orchestrator: dispatch one task to many agents, judge the results, ship the winner.","icon":"$undefined","url":"/docs/providers/moonshot-kimi","$ref":"providers/moonshot-kimi.md"},{"$id":"providers/minimax.md","type":"page","name":"MiniMax (direct API)","description":"Fact-checked 2026-06-10 against https://platform.minimax.io/docs/api-reference/text-openai-api.","icon":"$undefined","url":"/docs/providers/minimax","$ref":"providers/minimax.md"},{"$id":"providers/command-code.md","type":"page","name":"Command Code (CLI + Provider API)","description":"Reference for Command Code (commandcode.ai): the cmd coding CLI and its OpenAI-compatible Provider API. Read this before any Command Code adapter, headroom or fleet work.","icon":"$undefined","url":"/docs/providers/command-code","$ref":"providers/command-code.md"},{"$id":"providers/ollama.md","type":"page","name":"Ollama (local models)","description":"Drop-in for the existing LoopAgentAdapter (openai SDK): swap baseURL, static apiKey. Empirically verified on v0.30.6:","icon":"$undefined","url":"/docs/providers/ollama","$ref":"providers/ollama.md"},{"$id":"providers/ollama-cloud.md","type":"page","name":"Ollama Cloud (hosted inference)","description":"Fact-checked 2026-06-27 against https://docs.ollama.com/cloud and https://ollama.com/blog/cloud-models; cross-referenced with https://fabiorehm.com/blog/2026/04/12/pi-ollama-cloud-api/ (hands-on...","icon":"$undefined","url":"/docs/providers/ollama-cloud","$ref":"providers/ollama-cloud.md"},{"$id":"providers/direct-api.md","type":"page","name":"Direct-API Providers (OpenAI-compatible key-loop)","description":"This page covers the OpenAI-compatible direct-key providers registered in Cyberine as loop--build build-fleet workers. Each uses LoopAgentAdapter and requires only an API key — no CLI install, no...","icon":"$undefined","url":"/docs/providers/direct-api","$ref":"providers/direct-api.md"},{"$id":"providers/cloud-native.md","type":"page","name":"Cloud-Native Providers (Bedrock, Vertex AI)","description":"These two workers use cloud-platform auth rather than a plain API key. None of them routes through the LiteLLM gateway; each has a dedicated adapter in src/adapters/.","icon":"$undefined","url":"/docs/providers/cloud-native","$ref":"providers/cloud-native.md"},{"$id":"providers/cli-agents.md","type":"page","name":"CLI Coding Agents — adapter integration reference","description":"Pre-crawled, fact-checked reference for adding headless CLI coding agents as Cyberine workers (CliAgentAdapter spawns the official binary non-interactively, parses stdout into a Result)....","icon":"$undefined","url":"/docs/providers/cli-agents","$ref":"providers/cli-agents.md"},{"$id":"providers/claude-code-cli.md","type":"page","name":"Claude Code CLI (headless)","description":"Default model: claude-sonnet-5-5 (CYBERINECLAUDEMODEL; was claude-sonnet-5 until 2026-10-01, kvhxs). Auth dir allow-listed by the sandbox: ~/.claude.","icon":"$undefined","url":"/docs/providers/claude-code-cli","$ref":"providers/claude-code-cli.md"},{"$id":"providers/codex-cli.md","type":"page","name":"OpenAI Codex CLI (headless)","description":"Model: CYBERINECODEXMODEL if set, else Codex's configured default (no -m added). Deliberately unpinned (kd7ax, 2026-09-09): Codex 0.153.4 made GPT-6-Astra the bundled default, and Cyberine keeps...","icon":"$undefined","url":"/docs/providers/codex-cli","$ref":"providers/codex-cli.md"},{"$id":"providers/gemini-cli.md","type":"page","name":"Google Gemini CLI (headless)","description":"Default model: gemini-2.5-flash (CYBERINEGEMINIMODEL / GEMINIMODEL). Auth dir allow-listed by the sandbox: ~/.gemini.","icon":"$undefined","url":"/docs/providers/gemini-cli","$ref":"providers/gemini-cli.md"},{"$id":"providers/opencode-cli.md","type":"page","name":"OpenCode CLI (anomalyco/opencode, headless)","description":"Verified 2026-08-25 (hl-t693). opencode has two independent billing surfaces, and only ONE of them died. Getting this wrong in either direction is expensive, so read both halves:","icon":"$undefined","url":"/docs/providers/opencode-cli","$ref":"providers/opencode-cli.md"},{"$id":"providers/opencode2-v2-beta.md","type":"page","name":"OpenCode 2.0 beta (opencode2) — integrated as opencode2-deepseek, currently retired","description":"v2 is unreachable from every surface CHANGELOG-WATCH.md's OpenCode row polls, which is the whole reason a user report (\"opencode ra version 2 roi\") beat the sweep to it:","icon":"$undefined","url":"/docs/providers/opencode2-v2-beta","$ref":"providers/opencode2-v2-beta.md"},{"$id":"providers/muse-cli.md","type":"page","name":"Muse Code (Meta) — headless muse exec","description":"Every claim below was checked against the installed binary or a real dispatch. Where a measurement contradicted the expectation, the measurement is what is recorded and the expectation is named so...","icon":"$undefined","url":"/docs/providers/muse-cli","$ref":"providers/muse-cli.md"},{"$id":"providers/mcp-2.0-audit.md","type":"page","name":"MCP Spec RC (2026-07-28) + SDK 2.0.0-beta — Audit (T639)","description":"Audit-only. No code changed, no dependency bumped. Scope: src/mcp.ts (MCP server — exposes the orchestrator) and src/adapters/McpToolAdapter.ts (MCP client — consumes arbitrary remote MCP...","icon":"$undefined","url":"/docs/providers/mcp-2.0-audit","$ref":"providers/mcp-2.0-audit.md"}],"$id":"providers","$ref":{"folder":"providers","meta":"providers/meta.json"},"icon":"$undefined"},{"type":"folder","name":"Design notes","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"design/website-master.md","type":"page","name":"Website MASTER — \"Run record\" (marketing) + \"Day Assembly / Night Approach\" (docs)","description":"Read this before any website work. Locked 2026-10-01 by the operator. It supersedes the Plasma home + \"Six Instruments\" routes MASTER, archived as website-master-20260925-plasma.md; the rejected...","icon":"$undefined","url":"/docs/design/website-master","$ref":"design/website-master.md"},{"$id":"design/website-spec.md","type":"page","name":"Website spec — implementation contract for \"Parallel Cut\"","description":"Companion to website-master.md, which owns the IDENTITY (palette, type, signature suite, motion vocabulary). This file owns everything MASTER deliberately does not carry: information architecture,...","icon":"$undefined","url":"/docs/design/website-spec","$ref":"design/website-spec.md"},{"$id":"design/website-theme-decision.md","type":"page","name":"Website theme decision - Dracula, dark-only; rebuild epic shelved","description":"Decided 2026-08-27 by the user, in one sitting, after the three-way color contradiction below was surfaced. This note exists so the question is not re-opened from scratch: the site's palette is...","icon":"$undefined","url":"/docs/design/website-theme-decision","$ref":"design/website-theme-decision.md"},{"$id":"design/website-master-20260925-plasma.md","type":"page","name":"ARCHIVED — Plasma home + \"Six Instruments\" routes (2026-08-27 to 2026-10-01)","description":"Read this before any website work. It replaces website-master-20260827-parallel-cut.md, which was rejected at prototype the same day it was written — read its post-mortem before proposing...","icon":"$undefined","url":"/docs/design/website-master-20260925-plasma","$ref":"design/website-master-20260925-plasma.md"},{"$id":"design/website-master-20260827-parallel-cut.md","type":"page","name":"ARCHIVED — \"Parallel Cut\" (rejected at prototype, 2026-08-27)","description":"The single source of truth for the Cyberine website's visual identity. Locked 2026-08-27 by the user via /nf-design-system heavy, after an independent three-way brainstorm (codex gpt-5.6-sol at...","icon":"$undefined","url":"/docs/design/website-master-20260827-parallel-cut","$ref":"design/website-master-20260827-parallel-cut.md"},{"$id":"design/persona.md","type":"page","name":"Persona — Cyberine","description":"Cyberine ships a single built-in persona, Cyberine, that gives the chat planner a consistent voice and identity. The persona is optional: set CYBERINEPERSONA=none to disable it and run with bare...","icon":"$undefined","url":"/docs/design/persona","$ref":"design/persona.md"},{"$id":"design/autonomy-loop.md","type":"page","name":"Autonomy Loop — T455 Design Note","description":"Status: design (T455). Locks the architecture for the AUTONOMY epic (T456-T466). No product code changed. Implementation tasks must treat the contracts here as fixed. Source cross-checked:...","icon":"$undefined","url":"/docs/design/autonomy-loop","$ref":"design/autonomy-loop.md"},{"$id":"design/loop-sync-dispatch-verify.md","type":"page","name":"Loop Synchronous Dispatch + Verify Seam — Design Note","description":"Status: shipped — T459 (beam-search forks) and T463 (non-code verify gate) are wired into the loop's syncDispatchTask path, see syncDispatchTask, forkBeamSearchCandidates and verifyNonCodeResult...","icon":"$undefined","url":"/docs/design/loop-sync-dispatch-verify","$ref":"design/loop-sync-dispatch-verify.md"},{"$id":"design/code-context-3act.md","type":"page","name":"Code-Context 3-Act Architecture — T440 Design","description":"Status: design (T440). Locks the architecture for the CODE-CONTEXT epic (T441-T454). No product code changed. Implementation tasks must treat the contracts here as fixed. Source cross-checked:...","icon":"$undefined","url":"/docs/design/code-context-3act","$ref":"design/code-context-3act.md"},{"$id":"design/long-term-memory.md","type":"page","name":"Long-term memory for Cyberine — survey + design","description":"Agent memory is organized into five cognitive categories — episodic (experiences situated in time), semantic (abstract facts), procedural (skills, policies), working (in-context manipulation),...","icon":"$undefined","url":"/docs/design/long-term-memory","$ref":"design/long-term-memory.md"},{"$id":"design/knowledge-store.md","type":"page","name":"Knowledge and plan store for Cyberine","description":"Shipped 2026-11-10 (tasks k6tbw, k8380, kgt0g, kcnvb, k493v, kx5xv, k9nhg, ksrat). This page describes what the code does now, read from the source at the time of writing. Where the original plans...","icon":"$undefined","url":"/docs/design/knowledge-store","$ref":"design/knowledge-store.md"},{"$id":"design/lsp-diagnostic-feedback-loop.md","type":"page","name":"LSP Diagnostic Feedback Loop into Adapters — T436 Design Note","description":"Status: shipped (T436) — see src/core/config.ts (diagFeedback, diagFeedbackInterval, diagSeverity, diagMaxErrors) and src/adapters/LoopAgentAdapter.ts (runDiagnostics, per-edit and end-of-loop...","icon":"$undefined","url":"/docs/design/lsp-diagnostic-feedback-loop","$ref":"design/lsp-diagnostic-feedback-loop.md"},{"$id":"design/fleet-config-design.md","type":"page","name":"Fleet Config — Design Draft","description":"fleet.json declares policy (who is on, which model for which task). It never holds API keys or secrets. Keys stay exactly where they are today:","icon":"$undefined","url":"/docs/design/fleet-config-design","$ref":"design/fleet-config-design.md"},{"$id":"design/tui-config-interaction-standard.md","type":"page","name":"TUI Config Interaction Standard — T418 Tier-1 Design Doc","description":"Status: design artifact only — no code built. Tiers 2-3 (shared primitives + migration + settings screen) are TTY+decision-gated and not included here. Source cross-checked: src/tui/,...","icon":"$undefined","url":"/docs/design/tui-config-interaction-standard","$ref":"design/tui-config-interaction-standard.md"},{"$id":"design/cli-flag-audit.md","type":"page","name":"CLI Flag Audit (T428)","description":"Inventory of every top-level subcommand and sub-subcommand, with --help/-h status before and after T428.","icon":"$undefined","url":"/docs/design/cli-flag-audit","$ref":"design/cli-flag-audit.md"},{"$id":"design/cli-streaming-output.md","type":"page","name":"CLI Streaming Output — T435 Design Note","description":"Status: shipped — Path 1 landed (T435). See src/orchestrator/Orchestrator.ts:144,374-375. Source cross-checked: src/commands/fleet.ts, src/cli.ts, src/orchestrator/run.ts,...","icon":"$undefined","url":"/docs/design/cli-streaming-output","$ref":"design/cli-streaming-output.md"},{"$id":"design/hook-matcher-anchoring.md","type":"page","name":"Hook Matcher: Anchoring vs Exact Match (T563)","description":"Claude Code v2.1.195 changed hook matchers to require an EXACT match for hyphenated identifiers (e.g. code-reviewer, mcpbrave-search). Previously CC's matcher ran an unanchored substring test, so...","icon":"$undefined","url":"/docs/design/hook-matcher-anchoring","$ref":"design/hook-matcher-anchoring.md"},{"$id":"design/bus-presence-auth-scope.md","type":"page","name":"Bus/Presence/MCP Auth Scope (T584)","description":"apiTokenIsValid (src/store/apiTokens.ts) is wired into exactly one place: the REST authorized() check in src/web/security.ts, which is itself skipped entirely when the web server is bound to...","icon":"$undefined","url":"/docs/design/bus-presence-auth-scope","$ref":"design/bus-presence-auth-scope.md"},{"$id":"design/contextq-integration.md","type":"page","name":"ContextQ Integration — Cyberine long-term memory on the Agent Memory Layer","description":"Status: design spec (drives implementation). Grounded against ContextQ as it ships today (shared-context-server, agent layer migration 0043) and Cyberine memory waves 1-4.","icon":"$undefined","url":"/docs/design/contextq-integration","$ref":"design/contextq-integration.md"},{"$id":"design/plugin-marketplace-decision.md","type":"page","name":"Plugin Marketplace / Discovery Registry — Build vs Defer (T542)","description":"This note decides ONE thing: whether Cyberine should build a hosted marketplace / discovery registry for plugins — a remote index a user can search/browse, with install-from-registry as a third...","icon":"$undefined","url":"/docs/design/plugin-marketplace-decision","$ref":"design/plugin-marketplace-decision.md"}],"$id":"design","$ref":{"folder":"design","meta":"design/meta.json"},"icon":"$undefined"},{"type":"folder","name":"Evaluation","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"eval/worker-profile-coverage.md","type":"page","name":"Worker-profile eval coverage (T678)","description":"Generated by npx tsx scripts/t678-profile-coverage.ts --write. Re-run it after any fleet dispatch to refresh; do not hand-edit.","icon":"$undefined","url":"/docs/eval/worker-profile-coverage","$ref":"eval/worker-profile-coverage.md"},{"$id":"eval/code-context-3act-eval.md","type":"page","name":"3-Act Code-Context Eval — T453","description":"Status: complete. Offline measurement only. No live LLM calls made. Harness: src/eval/contextEval.ts Design doc: docs/design/code-context-3act.md","icon":"$undefined","url":"/docs/eval/code-context-3act-eval","$ref":"eval/code-context-3act-eval.md"}],"$id":"eval","$ref":{"folder":"eval","meta":"eval/meta.json"},"icon":"$undefined"}],"fallback":{"type":"root","$ref":{"folder":"","meta":"$undefined"},"$id":"fallback:root","name":"Docs","description":"$undefined","children":[{"$id":"fallback:privacy.md","type":"page","name":"Privacy and licence","description":"Source reviewed: 2026-11-08. This note covers the Cyberine application and plugin, not the retention policies of third-party services or a website hosting provider.","icon":"$undefined","url":"/docs/privacy","$ref":"privacy.md"},{"type":"folder","name":"Design","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"fallback:design/board-as-sole-coordination-channel.md","type":"page","name":"The board is the only coordination channel","description":"Operator ruling, 2026-09-21. Binding on every design that touches how two Cyberine sessions divide work.","icon":"$undefined","url":"/docs/design/board-as-sole-coordination-channel","$ref":"design/board-as-sole-coordination-channel.md"},{"$id":"fallback:design/claude-accounts.md","type":"page","name":"Shared Claude accounts: claude-accounts/1","description":"Status: accepted by the operator 2026-10-14 (relayed by cc-session-float28), with D2 changed to a single root variable. cc-session-float implements every NEW item as of its main 5c6b379....","icon":"$undefined","url":"/docs/design/claude-accounts","$ref":"design/claude-accounts.md"},{"$id":"fallback:design/fanout-topology.md","type":"page","name":"Fan-out topology audit (kkxs4)","description":"Evidence: Google Research, \"Towards a science of scaling agent systems\" (2026-01-28). Independent multi-agent systems amplified errors 17.2x and centralised ones 4.4x. On sequential tasks...","icon":"$undefined","url":"/docs/design/fanout-topology","$ref":"design/fanout-topology.md"},{"$id":"fallback:design/guard-fail-closed.md","type":"page","name":"Board guard: fail open, not onFailure: \"block\" (kjtjc)","description":"Rejected, 2026-10-10. The plugin's PreToolUse board guard stays fail-open: plugin/hooks/hooks.json does not set onFailure: \"block\", and plugin/hooks/board-hooks.mjs keeps mapping every CLI failure...","icon":"$undefined","url":"/docs/design/guard-fail-closed","$ref":"design/guard-fail-closed.md"},{"$id":"fallback:design/handoff.md","type":"page","name":"Shared session handoff: handoff/1","description":"Status: version 1, published 2026-10-17 by Cyberine (ai-agent-solution145) at the operator's word (D1-A), answering cc-session-float's request on its task k373d (operator option D: one spec,...","icon":"$undefined","url":"/docs/design/handoff","$ref":"design/handoff.md"},{"$id":"fallback:design/harness-contract.md","type":"page","name":"Harness contract: one core, thin shells","description":"Status: accepted with edits, re-verified against the source 2026-10-28 (task ktagh)","icon":"$undefined","url":"/docs/design/harness-contract","$ref":"design/harness-contract.md"},{"$id":"fallback:design/tui-direction-a.md","type":"page","name":"TUI direction A: layout contract","description":"Status: as-built contract, re-verified against the source 2026-10-28 (task kqk3m) and updated 2026-10-29 (kk04j, kh72a, ks08k). It states what the source does now, per view and per breakpoint;...","icon":"$undefined","url":"/docs/design/tui-direction-a","$ref":"design/tui-direction-a.md"},{"$id":"fallback:design/tui-hub-blocks.md","type":"page","name":"TUI hub: blocks, inspector and motion","description":"Status: shipped 2026-10-04 (kp3fe, k2k49, kf366, ky347, k90w4, kjge7; documented by k1kzw). This is the hub section of the TUI design for the direction the operator picked from the 2026-10-04...","icon":"$undefined","url":"/docs/design/tui-hub-blocks","$ref":"design/tui-hub-blocks.md"},{"$id":"fallback:design/tui-keymap-architecture.md","type":"page","name":"TUI keymap architecture (KEYMAP-V2)","description":"Status: design, 2026-09-02. Derived from a read of two shipped harnesses (Claude Code 2.1.252, Codex 0.149.1) plus this repo's own web dashboard, which already ships the pattern the TUI lacks.","icon":"$undefined","url":"/docs/design/tui-keymap-architecture","$ref":"design/tui-keymap-architecture.md"}],"$id":"fallback:design","$ref":{"folder":"design","meta":"$undefined"},"icon":"$undefined"},{"type":"folder","name":"Providers","root":"$undefined","defaultOpen":"$undefined","description":"$undefined","collapsible":"$undefined","children":[{"$id":"fallback:providers/cursor-agent.md","type":"page","name":"Cursor Agent CLI (headless)","description":"Binary is agent (the cursor-agent name is a back-compat alias). --force is mandatory for the CLI to actually apply file edits rather than only propose them; -p is the non-interactive/headless...","icon":"$undefined","url":"/docs/providers/cursor-agent","$ref":"providers/cursor-agent.md"}],"$id":"fallback:providers","$ref":{"folder":"providers","meta":"$undefined"},"icon":"$undefined"}]}},"tabs":[],"tabMode":"$undefined","sidebar":{"footer":"$L10"},"nav":{"title":"$L11"},"themeSwitch":{"enabled":false},"children":"$L12"}]}]]}]]}],"p":false,"v":null},"c":"$Q13"}]] 20:I[22530,["$2","$7","$8","$9","$a","$b","$c"],"ThemeToggle"] 22:I[97367,["$2"],"OutletBoundary"] 24:"/_next/static/chunks/2mrgr9w3cnjpj.js" 25:I[53266,["$2","$7","$8","$9","$a","$b","$c","$24"],"DocsPage"] 26:I[53266,["$2","$7","$8","$9","$a","$b","$c","$24"],"DocsTitle"] 27:I[53266,["$2","$7","$8","$9","$a","$b","$c","$24"],"DocsBody"] 28:I[27201,["$2"],"IconMark"] </article></body></html>