10 lines
84 KiB
JSON
10 lines
84 KiB
JSON
{"type":"session","version":3,"id":"019f9310-ee1e-7e02-8adc-277ee17bddc0","timestamp":"2026-07-24T07:39:48.126Z","cwd":"/home/liph/dotfiles/pi/.pi/agent"}
|
||
{"type":"model_change","id":"d55ecc00","parentId":null,"timestamp":"2026-07-24T07:39:49.198Z","provider":"ollama","modelId":"kimi-k2.7-code:cloud"}
|
||
{"type":"thinking_level_change","id":"0def96e7","parentId":"d55ecc00","timestamp":"2026-07-24T07:39:49.198Z","thinkingLevel":"medium"}
|
||
{"type":"message","id":"0429a573","parentId":"0def96e7","timestamp":"2026-07-24T07:40:04.721Z","message":{"role":"user","content":[{"type":"text","text":"<skill name=\"pi-subagents\" location=\"/home/liph/.pi/agent/npm/node_modules/pi-subagents/skills/pi-subagents/SKILL.md\">\nReferences are relative to /home/liph/.pi/agent/npm/node_modules/pi-subagents/skills/pi-subagents.\n\n# Pi Subagents\n\nThis skill is for the main parent orchestrator only. Do not inject or follow it inside spawned child subagents. The parent session owns delegation, orchestration, review fanout, and final fix-worker launches; child subagents should receive concrete role-specific tasks. Ordinary children should not run their own subagent workflows; the explicit exception is a delegated fanout child whose resolved builtin `tools` includes `subagent`, and that child may use `subagent` only for the fanout work the parent assigned.\n\nUse this skill when the parent orchestrator needs to launch a specialized subagent, compose multiple agents into a workflow, or create/edit agents and chains on demand.\n\n## When to Use\n\n- **Complex work orchestration**: use Fable mode as the default parent-agent loop for complex work. Complex means the task has multiple moving parts, unclear acceptance, cross-cutting code, meaningful user-visible impact, expensive or irreversible validation, broad review surface, or the user asks for orchestration. Lightweight one-off delegation can stay lightweight.\n- **Advisory review**: use fresh-context `reviewer` agents for adversarial code review, or fork to `oracle` when inherited decisions and drift matter\n- **Implementation handoff**: have `oracle` advise, then `worker` implement only after an approved direction\n- **Recon and planning**: use `scout` or `context-builder`, then `planner`\n- **Parallel exploration**: run multiple non-conflicting tasks concurrently\n- **Regular skill specialists**: when discovery shows proactive skill subagent suggestions and the current work is broad enough, launch a small fresh-context fanout that asks one subagent per relevant regularly used skill to apply that skill's perspective to the task\n- **Long-running work**: launch async/background runs and inspect them later. For mutation-capable work, bound the delivery slice and elapsed runtime, then request checkpoints after active tool work returns. Reserve hard turn and tool-call caps for explicitly read-only children.\n- **Subagent control**: watch needs-attention signals and soft-interrupt only when a delegated run is genuinely blocked\n- **Agent authoring**: create, update, or override agents and chains for a project\n\n## Tool vs Slash Commands\n\nAgents can use the `subagent(...)` tool directly for execution, management, status, and control.\nHumans often use the slash-command layer instead:\n\n- `/run` — launch a single agent\n- `/chain` — launch a chain of steps\n- `/parallel` — launch top-level parallel tasks\n- `/run-chain` — launch a saved `.chain.md` or `.chain.json` workflow\n- `/subagent-cost` — show parent plus child token usage and cost for the session\n- `/subagents-fleet` — open the live, inspection-only foreground/async fleet; `Ctrl+Alt+F` opens it during an active foreground turn, `↑↓`/`jk` selects children, and `PgUp`/`PgDn` scrolls transcript detail\n- `/subagents-doctor` — diagnose setup, discovery, async paths, and intercom bridge state\n- `/subagents-models [agent]` — show the live runtime-loaded builtin model mapping\n- `/subagents-profiles`, `/subagents-load-profile`, `/subagents-refresh-provider-models`, `/subagents-generate-profiles`, `/subagents-check-profile` — manage model profiles and provider catalogs\n- `/prompt-workflow` and `/chain-prompts` — run prompt templates through native subagent single/chain workflows\n\nPrefer the tool when you are writing agent logic. Prefer the slash commands when\nyou are guiding a human through an interactive flow.\n\nPackaged prompt shortcuts are also available for repeatable workflows. Treat them as reusable orchestration recipes, not just human slash commands. When the user asks for one of these shapes, or when the workflow clearly fits, apply the same pattern directly with `subagent(...)` and other tools:\n- `/parallel-review` — fresh-context reviewers with distinct review angles, then synthesis\n- `/review-loop` — parent-orchestrated worker, fresh-reviewer, and fix-worker cycles until clean or capped\n- `/parallel-research` — combine `researcher` and `scout` for external evidence plus local code context\n- `/parallel-context-build` — parallel `context-builder` passes that produce planning handoff context and meta-prompts\n- `/parallel-handoff-plan` — external-reference research plus local `context-builder` passes, followed by a synthesis handoff plan and implementation-ready meta-prompt\n- `/gather-context-and-clarify` — scout/research first, then ask the user clarifying questions with `interview`\n- `/parallel-cleanup` — two fresh-context reviewers (deslop + verbosity passes) for an adversarial cleanup review of the current diff\n\n## Applying Prompt Techniques Without Slash Commands\n\nThe prompt templates in `prompts/` encode workflows the parent agent can run on demand. If the user provides a URL, issue, PR, plan, local file, screenshot, or freeform target, treat that target as the primary scope: read or fetch it before launching children, then include it explicitly in every child task. Do not depend on the parent conversation history when the recipe calls for fresh context.\n\n### Parallel review technique\n\nUse this when the user wants adversarial review of a diff, plan, issue, file, or implemented work. Launch fresh-context `reviewer` agents with distinct angles generated from the actual target. Common angles are correctness/regressions, tests/validation, and simplicity/maintainability; adapt for TypeScript, UI, security, docs, or large structural changes. Reviewers should inspect files and diffs directly, return concise evidence-backed findings with file/line references, and avoid edits unless the user explicitly asks for a writer pass. The parent synthesizes fixes worth doing now, optional improvements, and feedback to ignore/defer before applying anything.\n\n### Proactive skill-specialist technique\n\nUse this when `{ action: \"list\" }` reports proactive skill subagent suggestions and the user's task would benefit from perspectives the parent regularly uses. These suggestions are conservative: a skill is recommended only when it is available and referenced repeatedly by configured agents or saved chains. Treat the list as an opt-in hint for the current task, not a command to always fan out.\n\nDefault guardrails:\n- Keep the fanout small: usually one or two skill-specialist children, never more than the listed recommendations or configured cap.\n- Prefer `context: \"fresh\"` and include only the files, diff, plan, URL, or request details each child needs. Use forked context only when private/session history is essential and appropriate to share.\n- Use read-only agents for analysis/review unless implementation was explicitly requested; do not create several writers in the same worktree.\n- Skip proactive skill subagents for tiny questions, direct commands, highly private requests, or when the user asks not to delegate.\n- Make cost and concurrency visible by using an ordinary `subagent(...)` call rather than hidden/background automation.\n\nExample shape:\n\n```typescript\nsubagent({\n tasks: [\n { agent: \"reviewer\", task: \"Apply the available 'deslop' skill to review the current diff for concrete cleanup findings only. Do not modify files.\", skill: \"deslop\" },\n { agent: \"reviewer\", task: \"Apply the available 'accessibility' skill to review the UI changes for concrete issues only. Do not modify files.\", skill: \"accessibility\" }\n ],\n context: \"fresh\",\n concurrency: 2\n})\n```\n\n### Review-loop technique\n\nUse this when the user wants implementation or current diff review to continue until reviewers stop finding fixes worth doing now. Keep the loop in the parent session: one async `worker` implements or fixes, fresh-context `reviewer` agents inspect the actual repo and diff, the parent synthesizes accepted fixes, and one async forked `worker` applies them. The parent can express the sequence up front as an async/background chain when the workflow is known, or continue with explicit follow-up subagent runs after each async completion. For an initial chain, pass `async: true` so the main chat is unblocked; do not set `clarify: true` unless the user explicitly wants the foreground clarify UI. Treat an async implementation worker handoff as an intermediate state, not final completion, unless the user explicitly asked for worker-only work, review-only output, or to stop after implementation. Stop when reviewers find no blockers or fixes worth doing now, remaining feedback is optional or deferred, an unapproved product/scope/architecture decision appears, or the max review-round cap is reached. Default to 3 review rounds unless the user sets a different cap. Do not loop for optional polish, and do not let children launch subagents or decide the loop outcome.\n\nAs a conservative orchestration policy, do not pass `turnBudget` or a hard `toolBudget` to an implementation worker, fix worker, reviewer with edit authority, or other mutation-capable child. The default tool budget blocks read/search tools rather than mutation tools, but count limits still do not measure delivery safety. Use a narrow task plus an outer elapsed deadline with enough margin, then request a checkpoint after the current tool returns. The checkpoint should report changed files, build/test state, remaining work, and commit or PR state. An elapsed timeout is not a mutation-safe boundary and must not be used as the checkpoint trigger.\n\n### Parallel research technique\n\nUse this when the question needs both external evidence and local implications. Combine `researcher` for official docs, specs, ecosystem behavior, recent changes, benchmarks, and primary sources with `scout` for repository files, patterns, constraints, tests, and likely integration points. Give each child a distinct angle: external evidence, local code context, and practical tradeoffs. Ask for source links or file ranges, confidence level, gaps, and decision implications. Do not ask these children to edit unless implementation was explicitly requested.\n\n### Parallel context-build technique\n\nUse this before planning or implementation when a stronger handoff is needed. Run a chain with one parallel step of `context-builder` agents rather than top-level parallel tasks, so relative output files live under the temporary chain directory. Give every task a distinct output path such as `context-build/request-and-scope.md`, `context-build/codebase-and-patterns.md`, and `context-build/validation-and-risks.md`. Choose two or three builders: request/scope, codebase/patterns, and validation/risks. Each builder must read every relevant file needed to understand its slice, follow imports/callers/tests/docs/config, conduct tool-available web research when needed, and include a compact `meta-prompt` section. The parent synthesizes the outputs into important context, recommended next meta-prompt, open questions, assumptions, and artifact paths.\n\nExample shape:\n\n```typescript\nsubagent({\n chain: [{\n parallel: [\n { agent: \"context-builder\", task: \"Build request/scope context for: ...\", output: \"context-build/request-and-scope.md\" },\n { agent: \"context-builder\", task: \"Build codebase/pattern context for: ...\", output: \"context-build/codebase-and-patterns.md\" },\n { agent: \"context-builder\", task: \"Build validation/risk context for: ...\", output: \"context-build/validation-and-risks.md\" }\n ]\n }],\n context: \"fresh\"\n})\n```\n\n### Parallel handoff-plan technique\n\nUse this when the user needs a solution brief or implementation-ready handoff from an external reference plus local code context, such as “study this library behavior, inspect our codebase, then produce a worker prompt.” Run a chain with a first parallel group and a second synthesis `context-builder` step. The first group usually includes `researcher` for external projects/docs/prompt guidance and `context-builder` for local code context; add a second `context-builder` for implementation strategy only when the scope is large enough to benefit. Use distinct output paths under `handoff/`, then have the synthesis `context-builder` read those outputs and write `handoff/final-handoff-plan.md` with the recommended approach, likely files, constraints, non-goals, validation, risks, unresolved questions, and final compact implementation-ready meta-prompt.\n\nExample shape:\n\n```typescript\nsubagent({\n chain: [\n { parallel: [\n { agent: \"researcher\", task: \"Research the external reference and transferable implementation ideas for: ...\", output: \"handoff/external-reference.md\" },\n { agent: \"context-builder\", task: \"Build local codebase context for: ...\", output: \"handoff/local-context.md\" },\n { agent: \"context-builder\", task: \"Compare evidence and propose implementation strategy for: ...\", output: \"handoff/implementation-strategy.md\" }\n ] },\n { agent: \"context-builder\", task: \"Read {previous} and synthesize the final handoff plan and implementation-ready meta-prompt.\", output: \"handoff/final-handoff-plan.md\" }\n ],\n context: \"fresh\"\n})\n```\n\n### Gather-context-and-clarify technique\n\nUse this at the start of non-trivial work. Launch `scout` for local context and `researcher` only when external docs, recent sources, ecosystem context, or primary evidence would materially improve understanding. Ask children for concise findings plus remaining clarification questions. Then synthesize what is known and use `interview` to ask the unresolved questions needed for shared understanding before planning or implementing.\n\n### Parallel cleanup technique\n\nUse this after implementation when the user wants cleanup review or when a final pass would reduce AI-slop. Launch two fresh-context `reviewer` tasks with `output: false` and `progress: false`: one deslop pass and one verbosity pass. If the `deslop` or `verbosity-cleaner` skills are available, pass the relevant skill to that reviewer; otherwise inline the criteria. Both reviewers are review-only and should flag concrete issues with severity, file/line references, and smallest safe fixes. Phrase the constraint as “Do not modify project/source files; returning findings through the configured output artifact is allowed” when you use `output` or `outputMode: \"file-only\"`. The parent decides what to apply and asks before making changes unless cleanup was already authorized.\n\n### Staged fix orchestration technique\n\nUse this when a broad diff has known reviewer findings across several items and the user wants the parent to “orchestrate subagents like a boss.” Keep the active worktree safe with a three-stage chain:\n\n1. A parallel read-only planning fanout, one planner/reviewer per issue cluster. Each child inspects the real diff and returns exact files, line refs, proposed fixes, and focused validation. They must not edit.\n2. One writer worker. It receives the planner summaries through `{previous}`, the parent’s accepted scope, stop rules, and verification contract. It is the only child allowed to edit the active worktree.\n3. A parallel read-only validation fanout. Validators inspect the worker diff from fresh context with distinct angles, report pass/fail, remaining blockers, and missing verification.\n\nPrefer `async: true`, `context: \"fresh\"` for planners/validators, `outputMode: \"file-only\"` for large summaries, and per-stage output names that will not collide. Add `phase` and `label` to make async status readable, and use `as` plus `{outputs.name}` when a later step needs a specific earlier result instead of the whole `{previous}` blob. Use this pattern instead of launching several writer workers into a dirty worktree. Include non-blocking suggestions in the writer prompt only when they are small, safe, and do not expand product scope; otherwise record them as deferred.\n\nWhen the first step can return a structured target list, prefer dynamic fanout instead of hand-authoring a static parallel group. Use `outputSchema` and `as` on the producer, then an `expand` step with `from: { output, path }`, an explicit `maxItems`, one `parallel` child template, and `collect.as`. Item templates may use `{item}` or a named item such as `{target.path}`. Do not use dynamic fanout for prose outputs, nested fanout, dynamic agent selection, reducers, `when` conditions, or arbitrary expressions; `.chain.md` does not support this syntax, so use direct JSON or a saved `.chain.json`.\n\nExample shape:\n\n```typescript\nsubagent({\n async: true,\n context: \"fresh\",\n chain: [\n { parallel: [\n { agent: \"reviewer\", phase: \"Planning\", label: \"Deploy docs\", as: \"deployPlan\", task: \"Plan fixes for deploy docs/workflow. Inspect the current diff. Do not modify project/source files; returning findings via the configured output artifact is allowed.\", output: \"plans/deploy.md\", outputMode: \"file-only\" },\n { agent: \"reviewer\", phase: \"Planning\", label: \"Scheduler contract\", as: \"schedulerPlan\", task: \"Plan fixes for scheduler contract. Inspect the current diff. Do not modify project/source files; returning findings via the configured output artifact is allowed.\", output: \"plans/scheduler.md\", outputMode: \"file-only\" },\n { agent: \"reviewer\", phase: \"Planning\", label: \"Sandbox/security\", as: \"sandboxPlan\", task: \"Plan fixes for sandbox/security. Inspect the current diff. Do not modify project/source files; returning findings via the configured output artifact is allowed.\", output: \"plans/sandbox.md\", outputMode: \"file-only\" }\n ], concurrency: 3 },\n { agent: \"worker\", phase: \"Implementation\", label: \"Apply accepted fixes\", as: \"workerResult\", task: \"Apply only the accepted fixes from these planning summaries. You are the sole writer for the active worktree. Run focused validation and report changed files, commands, failures, and remaining issues.\\n\\nDeploy plan:\\n{outputs.deployPlan}\\n\\nScheduler plan:\\n{outputs.schedulerPlan}\\n\\nSandbox plan:\\n{outputs.sandboxPlan}\", output: \"worker/fixes.md\", outputMode: \"file-only\", progress: true },\n { parallel: [\n { agent: \"reviewer\", phase: \"Validation\", label: \"Deploy/scheduler validation\", task: \"Validate the post-worker diff for deploy and scheduler fixes. Start from the worker result: {outputs.workerResult}. Do not modify project/source files; returning findings via the configured output artifact is allowed.\", output: \"validation/deploy-scheduler.md\", outputMode: \"file-only\" },\n { agent: \"reviewer\", phase: \"Validation\", label: \"Sandbox validation\", task: \"Validate the post-worker diff for sandbox/security fixes. Start from the worker result: {outputs.workerResult}. Do not modify project/source files; returning findings via the configured output artifact is allowed.\", output: \"validation/sandbox.md\", outputMode: \"file-only\" }\n ], concurrency: 2 }\n ]\n})\n```\n\n## Builtin Agents\n\nBuiltin agents load at the lowest priority. Project agents override user agents,\nand user/project agents override builtins with the same name.\n\n| Agent | Purpose | Model | Typical output / role |\n|-------|---------|-------|------------------------|\n| `scout` | Fast codebase recon | inherits default | Writes `context.md` handoff material |\n| `planner` | Creates implementation plans | inherits default | Writes `plan.md` |\n| `worker` | Implementation and approved oracle handoffs | inherits default | Single-writer implementation with decision escalation |\n| `reviewer` | Review-and-fix specialist | inherits default | Can edit/fix reviewed code |\n| `context-builder` | Requirements/codebase handoff builder | inherits default | Writes structured context files |\n| `researcher` | Web research brief generator | inherits default | Writes `research.md` |\n| `delegate` | Lightweight generic delegate | inherits default | No fixed output; generic delegated work |\n| `oracle` | Decision-consistency advisory review | inherits default | Advisory review, intercom coordination |\n\nBuiltin agents inherit the current Pi default model unless a run, user setting, project setting, or `subagents.defaultModel` overrides `model`. Set `subagents.defaultModel` when subagents should use a different default model than the parent session. Override builtin defaults before copying full agent files when a small tweak is enough.\n\nFor one run, use inline config:\n\n```text\n/run reviewer[model=anthropic/claude-sonnet-4] \"Review this diff\"\n```\n\nFor persistent tweaks, edit `subagents.agentOverrides` in user or project settings. User overrides apply everywhere. Project overrides apply only in that repo and win over user overrides. Use `/subagents-models` or `subagent({ action: \"models\" })` to inspect the live mapping after settings and overrides load.\n\nModel ids do not have to be exact. Separator variations (`claude-haiku-4.5` vs `claude-haiku-4-5`), case (`Claude-Sonnet-4`), and optional trailing date stamps (`claude-haiku-4-5-20251001`) all resolve to the same registry model. Exact `provider/id` wins; a qualified `provider/model` never switches providers. To constrain subagents to a budget or compliance profile, set `subagents.modelScope: { enforce: true, allow: [\"anthropic/*\", \"openai/gpt-5-*\"] }` in user or project settings. Out-of-scope models you pass explicitly error and abort; models inherited from frontmatter, `subagents.defaultModel`, agent frontmatter, or the parent session only warn.\n\nFor model fleets, use the profile commands instead of hand-editing repeated overrides: `/subagents-refresh-provider-models <provider>`, `/subagents-generate-profiles <provider>`, `/subagents-load-profile <name>`, and `/subagents-check-profile <name>`. Profiles live under `~/.pi/agent/profiles/pi-subagents/` and replace only `settings.subagents` when loaded.\n\n## Prompting role subagents\n\nBuiltin role agents inherit the current Pi default model unless you override them. When launching them, write the task prompt as a compact contract, not a long procedural script. Define the destination and let the role choose the efficient path.\n\nA strong subagent prompt usually includes:\n- **Goal**: the concrete outcome the child should produce.\n- **Context/evidence**: relevant plan paths, files, diffs, decisions, or user constraints already approved.\n- **Success criteria**: what must be true before the child can finish.\n- **Hard constraints**: true invariants only, such as no edits for review-only tasks, one writer thread, child must not run subagents unless it is an explicitly assigned `tools: subagent` fanout child, or escalation for unapproved decisions.\n- **Validation**: targeted checks to run, or the next-best check when validation is impossible.\n- **Output**: the expected summary shape, artifact path, or finding format.\n- **Stop rules**: when to ask via `intercom`, when to stop after enough evidence, and when not to keep searching.\n\nAvoid carrying over old prompt habits that over-specify every step. Use `must`, `always`, and `never` for real invariants; for judgment calls, give decision rules. For example, tell a reviewer to inspect the staged diff directly and report only evidence-backed findings, rather than prescribing every file or command. Tell a researcher the retrieval budget: start with broad targeted searches, fetch only the strongest sources, search again only when a required fact is missing, then stop.\n\nFor implementation handoffs, name the approved scope and success criteria more clearly than the process. Good prompts say what to change, what not to change, where the evidence lives, how to validate, and when to escalate. They should not ask the child to create another subagent plan or continue the parent conversation.\n\nSettings locations:\n- User scope: `~/.pi/agent/settings.json`\n- Project scope: `.pi/settings.json`\n\nDirect settings example:\n\n```json\n{\n \"subagents\": {\n \"agentOverrides\": {\n \"reviewer\": {\n \"model\": \"anthropic/claude-sonnet-4\",\n \"thinking\": \"high\",\n \"fallbackModels\": [\"openai/gpt-5-mini\"],\n \"acceptanceRole\": \"read-only\"\n }\n }\n }\n}\n```\n\nUseful override fields: `model`, `fallbackModels`, `thinking`,\n`systemPromptMode`, `inheritProjectContext`, `inheritSkills`, `defaultContext`,\n`acceptanceRole`, `disabled`, `skills`, `tools`, and `systemPrompt`. Use\n`acceptanceRole: false` to clear an override. Create a user or project\nagent with the same name only when you want a substantially different agent.\n\nIf a provider rejects model IDs with thinking suffixes, use\n`subagents.disableThinking: true` in user or project settings to clear bundled\nbuiltin thinking defaults globally. A higher-precedence per-agent `thinking`\noverride can opt one builtin back in.\n\nTool description modes live in `~/.pi/agent/extensions/subagent/config.json`, not `subagents` settings. Set `toolDescriptionMode` to `compact` to reduce tool-description prompt cost while keeping the execution, async/`subagent_wait`, child-safety, one-writer, management/action, and artifact/status guardrails. Set it to `custom` to read `subagent-tool-description.md` from the project config dir or agent dir; invalid custom files fall back to full mode and the safety guidance is still appended.\n\n## Discovery and Scope Rules\n\nAgent files can live in:\n- `~/.pi/agent/agents/**/*.md` — user scope\n- `.pi/agents/**/*.md` — canonical project scope\n- legacy `.agents/**/*.md` — still read for compatibility, but `.pi/agents/` wins on conflicts\n\nChains live in:\n- `~/.pi/agent/chains/**/*.chain.md` and `~/.pi/agent/chains/**/*.chain.json` — user scope\n- `.pi/chains/**/*.chain.md` and `.pi/chains/**/*.chain.json` — project scope\n\nDiscovery is recursive. `.chain.md` files do not define agents. Use `.chain.md` for simple saved chains and `.chain.json` for dynamic fanout or inline schema objects. Agents and chains can set optional frontmatter/package metadata; `name: scout` plus `package: code-analysis` registers as runtime name `code-analysis.scout` while serialization keeps `name` and `package` separate.\n\nPrecedence is by parsed runtime name:\n1. project scope\n2. user scope\n3. builtin agents\n\n## Running Subagents\n\n### Single agent\n\n```typescript\nsubagent({\n agent: \"oracle\",\n task: \"Review my current direction and challenge assumptions.\"\n})\n```\n\n### Forked context\n\n```typescript\nsubagent({\n agent: \"oracle\",\n task: \"Review my current direction and challenge assumptions.\"\n})\n```\n\n`context: \"fork\"` creates a branched child session from the current persisted\nparent session. It does **not** create a fresh minimal review context or filter\nhistory down to only the relevant parts. Use it when you want a separate review\nor execution thread that can still reference the parent session history.\n\n### Parallel execution\n\n```typescript\nsubagent({\n tasks: [\n { agent: \"scout\", task: \"Explore the auth module\" },\n { agent: \"reviewer\", task: \"Review the API client\" }\n ]\n})\n```\n\nTop-level parallel tasks can override per-task behavior:\n\n```typescript\nsubagent({\n tasks: [\n { agent: \"scout\", task: \"Map auth\", output: \"auth-context.md\", progress: true },\n { agent: \"researcher\", task: \"Research OAuth best practices\", output: \"oauth-research.md\" },\n { agent: \"reviewer\", task: \"Review auth tests\", model: \"anthropic/claude-sonnet-4\" }\n ],\n concurrency: 3\n})\n```\n\nAvoid duplicate output paths in parallel tasks. Concurrent children should not write to the same file. For large saved outputs, set `outputMode: \"file-only\"` together with an `output` path. The parent result then contains only a compact reference like `Output saved to: /abs/report.md (48.2 KB, 2847 lines). Read this file if needed.` instead of the full saved content. Do not use `output: false` for this; `output: false` means no file output. Read-only children return the complete artifact in their final response and the runtime persists it, so missing write tools are not a supervisor blocker. Mutation-capable children still receive direct-write instructions. Failed runs and save errors still return inline details for debugging.\n\n### Chain execution\n\n```typescript\nsubagent({\n chain: [\n { agent: \"scout\", task: \"Map the auth flow and summarize key files\" },\n { agent: \"planner\", task: \"Create an implementation plan from {previous}\" },\n { agent: \"worker\", task: \"Implement the approved plan based on {previous}\" }\n ]\n})\n```\n\nChain steps can use templated variables such as `{task}`, `{previous}`,\n`{chain_dir}`, and `{outputs.name}`. Use `as: \"name\"` on a successful step or\nparallel task to make that output available to later steps. Prefer named outputs\nwhen a later step needs one specific result; keep `{previous}` for simple linear\nhandoffs or full fan-in summaries. Use `phase` and `label` for status readability.\nUse `outputSchema` when later steps need reliable structured data; the child must\ncall `structured_output` with schema-valid JSON, or the step fails.\n\n### Async/background\n\nPrefer async mode for every subagent launch. Set `async: true` no matter the task unless there is a specific reason to opt into a foreground/blocking run. This applies to scouts, researchers, workers, reviewers, validators, oracle checks, one-off delegates, chains, and parallel groups. Keep the write path single-threaded even when the run is async.\n\nAsync does not mean parallel writes. Do not edit the same active worktree while an async worker is changing it. Parent-side overlap should be reading, validation prep, synthesis, command planning, or review of unaffected context unless the writer is isolated in a separate worktree.\n\nDo not end your turn immediately after launching an async child if you promised to keep working. Continue the local inspection, synthesis, or validation prep, then check the async run when its result is needed.\n\nIn an interactive chat, normally return control when ready to yield and let Pi wake the session on completion; do not call `subagent_wait()` merely to wait. Override that default and call it when the current request is run-to-completion — for example, the user asked you to report results back before continuing or a skill cannot return before its background work finishes. Headless sessions auto-drain exact current-session work at `agent_end`; call `subagent_wait()` when this turn must receive results before it ends. Never substitute sleep or status-polling loops.\n\n`subagent_wait()` returns when the next initially active async run or registered provider item finishes or a subagent needs attention. Use `subagent_wait({ all: true })` for all work active at call time, `subagent_wait({ id: \"...\" })` for one async or remembered detached foreground run, and `subagent_wait({ timeoutMs })` to cap the block. If a foreground child detaches for supervisor coordination, reply first, then wait on its id; do not resume or launch a replacement while it remains detached. Headless sessions also auto-drain exact current-session work at `agent_end` as a final safeguard.\n\n```typescript\nsubagent({\n agent: \"worker\",\n task: \"Run the full test suite\",\n async: true\n})\n```\n\nFile-only output mode also works for async single runs, top-level parallel task items, sequential chain steps, and chain parallel task items. In chains, `{previous}` receives the compact saved-file reference when the prior step used file-only mode.\n\nFor review fanout where the parent continues a local audit:\n\n```typescript\nconst run = subagent({\n agent: \"reviewer\",\n task: \"Review the current diff for correctness issues. Do not edit files.\",\n async: true,\n context: \"fresh\"\n})\n// Continue local inspection, then later call status with the returned id.\n```\n\nInspect async runs with `subagent({ action: \"status\", id: \"...\" })` or `subagent({ action: \"status\" })` for active runs. Use `subagent({ action: \"status\", view: \"fleet\" })` when supervising several active foreground/background runs and `subagent({ action: \"status\", id: \"...\", view: \"transcript\", index: 0 })` when you need the latest child output without digging through artifacts. If a delegated fanout child launches nested runs, the parent status view shows them as a tree and you can target a nested run directly with its nested id.\n\nUse `steer` for top-level live async guidance and `resume` after a delegated run pauses or finishes. Routed nested runs retain their existing non-destructive live follow-up path:\n\n```typescript\nsubagent({ action: \"steer\", id: \"run-id\", message: \"Focus on the failing test.\" })\nsubagent({ action: \"resume\", id: \"run-id\", message: \"Follow up on this point.\" })\nsubagent({ action: \"resume\", id: \"run-id\", index: 1, message: \"Continue reviewer 2.\" })\nsubagent({ action: \"resume\", id: \"nested-run-id\", message: \"Continue this nested reviewer.\" })\n```\n\nResume behavior:\n- `resume` revives paused, completed, or failed async/foreground children from persisted session files; stopped runs remain non-resumable, and it does not interrupt live top-level async children.\n- Use `steer` for acknowledged guidance to a live top-level async child.\n- A live nested run can still receive a non-destructive `resume` follow-up through its owner route.\n- If an async child has completed, `resume` revives it by starting a new async child from the persisted child session file.\n- Multi-child async runs require `index` unless only one running child is selectable.\n- Completed foreground single, parallel, and chain runs can also be revived by `index` while their run metadata remains in extension state.\n- Nested runs can be resumed by nested id when a live route or persisted nested session metadata is available.\n- Revive starts a new child process from the old session context; it does not restart the same OS process.\n- Direct revival holds an exclusive cross-process lease on the canonical child session file until the new child finishes. Concurrent attempts fail before Pi starts and identify the owning revived run; stale ownership is reclaimed only when the recorded process is demonstrably gone or reused.\n- If the chosen child has no persisted `.jsonl` session file, resume fails and reports that directly.\n\nUse diagnostics when setup or child startup looks wrong:\n\n```typescript\nsubagent({ action: \"doctor\" })\n```\n\n### Scheduled subagent runs\n\nScheduled runs defer a subagent launch until a future time. They are opt-in and require `{ \"scheduledRuns\": { \"enabled\": true } }` in `~/.pi/agent/extensions/subagent/config.json`. Only schedule explicit delayed runs the user asked for; do not schedule runs speculatively.\n\n```typescript\n// Launch a reviewer in 30 minutes\nsubagent({ action: \"schedule\", agent: \"reviewer\", task: \"Review the diff for correctness issues.\", schedule: \"+30m\", scheduleName: \"evening review\" })\n\n// Schedule a parallel fanout\nsubagent({ action: \"schedule\", tasks: [{ agent: \"scout\", task: \"Map the auth module\" }, { agent: \"scout\", task: \"Map the billing module\" }], schedule: \"+1h\" })\n\n// Inspect, list, and cancel\nsubagent({ action: \"schedule-list\" })\nsubagent({ action: \"schedule-status\", id: \"ab12\" })\nsubagent({ action: \"schedule-cancel\", id: \"ab12\" })\n```\n\n`schedule` accepts the same execution fields as a normal async run (`agent`/`tasks`/`chain`, `cwd`, `model`, `output`, `reads`, `progress`, `acceptance`, `timeoutMs`) plus `schedule` (a relative delay like `+10m`/`+2h`/`+1d` or a future ISO timestamp with a timezone such as `2030-01-01T09:00:00Z`) and an optional `scheduleName`. Scheduled runs always launch async with fresh context; `context: \"fork\"`, `async: false`, and `clarify: true` are rejected. Once the timer fires, the run becomes a normal tracked async run: it appears in the async widget, is inspectable with `subagent({ action: \"status\" })`, can be awaited with `subagent_wait()`, and delivers the normal completion notification.\n\nSchedules are persisted per session and restored after a Pi restart. A job whose scheduled time passed by more than `scheduledRuns.maxLatenessMs` (default 5 minutes) while Pi was unavailable is marked `missed` instead of firing late. `scheduledRuns.maxPending` (default 20) caps pending or running scheduled jobs per session.\n\nHumans can use `/subagents-doctor` for the same read-only report. It checks runtime paths, discovery counts, async support, current session context, and intercom bridge state.\n\n### Subagent control\n\nSubagent control is the runtime visibility and intervention layer for delegated runs. It is separate from lifecycle status. Lifecycle status says whether a child is `queued`, `running`, `paused`, `complete`, or `failed`. Activity reporting is factual: it tracks the last observed activity time and the current tool when known. It does not pretend to know that a child is truly stuck.\n\nDefault behavior is intentionally conservative. When no activity has been observed past the configured threshold, the run emits a `needs_attention` control event. Foreground runs can push this as a `subagent:control-event` event, and async runs persist it to `events.jsonl` so the parent tracker can surface it without constant manual polling. Notification-worthy control events are also inserted into the visible transcript so both the user and the parent agent can see them, with a proactive hint plus concrete `nudge`, `status`, and `interrupt` options. Visible notifications fire once per child run and attention state.\n\nUse soft interrupt when a child is clearly blocked or drifting and the parent needs to regain control:\n\n```typescript\nsubagent({ action: \"interrupt\" })\n```\n\nPass `id` when targeting a specific controllable run, including a nested run shown in the parent status tree:\n\n```typescript\nsubagent({ action: \"interrupt\", id: \"abc123\" })\nsubagent({ action: \"interrupt\", id: \"nested-run-id\" })\n```\n\nA soft interrupt cancels the current child turn and leaves the run paused. It does not mean the delegated task succeeded or failed. Bare `interrupt` does not target hidden nested descendants; use the explicit nested id. After an interrupt, decide the next explicit action: resume with clearer instructions, replace the task, ask the user, or stop the workflow.\n\nPer-run control thresholds can be overridden when a task legitimately runs without observable output for longer than usual:\n\n```typescript\nsubagent({\n agent: \"worker\",\n task: \"Run the slow migration test suite\",\n control: {\n needsAttentionAfterMs: 300000,\n notifyOn: [\"needs_attention\"]\n }\n})\n```\n\nIf the run already has an active intercom bridge target, needs-attention notifications can also prepare a compact intercom ping for the orchestrator. When a child route is available, the ping tells the orchestrator which agent needs attention and includes the exact `intercom({ action: \"send\", to: \"...\" })` target for a nudge. Do not invent a target or ask the child to self-report when no bridge exists.\n\nSteering is acknowledged delivery, not a send attempt or model-compliance signal:\n\n```typescript\nsubagent({ action: \"steer\", id: \"abc123\", message: \"Focus on the failing test.\" })\n```\n\nThe action waits up to three seconds for the child Pi session to accept the correlated user input and returns a request id with `delivered`, `scheduled`, `pending`, `partial`, `recovered`, or `failed` plus per-child states. Indexed pending children return `scheduled` immediately. Only a top-level single-child run may automatically interrupt after a missed acknowledgment and recover after confirmed pause within a further 15 seconds. Recovery preserves the original child contract and only its remaining deadline, turn, and tool budgets. If the session is missing, a budget is exhausted, the pause cannot be confirmed, or replacement launch fails, the source remains paused when pausing succeeded and the action returns the exact failure. Chain, parallel, and nested runs never auto-interrupt; inspect their per-child outcomes and handle failures explicitly. A late acknowledgment is recorded and cannot cancel committed recovery.\n\n## Clarify TUI\n\nSingle and parallel runs support a clarification TUI when you want to preview or\nedit parameters before launch:\n\n```typescript\nsubagent({\n agent: \"worker\",\n task: \"Implement feature X\",\n clarify: true\n})\n```\n\nTool calls launch directly by default. Set `clarify: true` on single, parallel, or chain runs when you want the clarify UI. Clarify edits affect only the next run; use management actions, settings, or markdown files for persistent changes.\nFor programmatic background launches, use `async: true`. `clarify: true` keeps the run foreground for the clarify UI.\n\n\n## Worktree Isolation\n\nWhen multiple agents might write concurrently, use worktrees instead of letting\nthem share one filesystem view.\n\n```typescript\nsubagent({\n tasks: [\n { agent: \"worker\", task: \"Implement feature A\" },\n { agent: \"worker\", task: \"Implement feature B\" }\n ],\n worktree: true\n})\n```\n\n`worktree: true` gives each parallel task its own git worktree branched from\nHEAD. This requires a clean git state and is mainly for intentionally parallel\nwrite workflows. If you want one writer thread and several advisory agents,\nprefer a single-writer pattern instead.\n\n## The Oracle Workflow\n\nThe intended oracle loop is:\n1. the main agent forks to `oracle`\n2. `oracle` reviews direction, drift, assumptions, and risks\n3. `oracle` can coordinate back through `contact_supervisor` when the bridge injects it\n4. the main agent decides what direction to approve\n5. only then should `worker` implement\n\n```typescript\n// Advisory review in a branched thread. Oracle defaults to forked context.\nsubagent({\n agent: \"oracle\",\n task: \"Review my current direction, challenge assumptions, and propose the best next move.\"\n})\n\n// Implementation only after explicit approval. Worker defaults to forked context.\nsubagent({\n agent: \"worker\",\n task: \"Implement the approved approach: ...\"\n})\n```\n\n`oracle` is not a fresh-context reviewer in the Cognition article sense. It is\na forked advisory thread that inherits the parent session history and uses that\nhistory as a baseline contract.\n\nUse `oracle` as a smart-friend escalation when the parent needs help with trajectory rather than diff inspection: architectural boundaries, model capability routing, merge conflicts, reviewer disagreement, context drift after long work, a worker about to invent a pattern, or fixes that require product/scope tradeoffs. Ask broad questions when the right concern is unclear, and let `oracle` point out missing context or files the parent should inspect before asking again. Keep `oracle` advisory unless it has been explicitly assigned the single writer role.\n\n## Subagent + Intercom Coordination\n\n`pi-subagents` includes native supervisor coordination. Child agents can use `contact_supervisor` to ask the exact parent session that spawned them; messages are scoped by parent session id and should not appear in other Pi sessions.\n\nMost agents should not call generic `intercom` directly unless bridge instructions provide a target and `contact_supervisor` is unavailable. Do not invent a target. Prefer the tool from the injected bridge instructions.\n\nUse `contact_supervisor` with `reason: \"need_decision\"` when:\n- a subagent is blocked on a decision\n- a child needs clarification instead of guessing\n- an approval, product, API, or scope choice is required before continuing safely\n\nUse `contact_supervisor` with `reason: \"interview_request\"` when the child needs structured supervisor input rather than a freeform answer. The request waits for a parent reply, so the child should stay alive and continue only after the reply arrives.\n\nDo not use `contact_supervisor` just to resolve review-only/no-project-edit versus progress-writing or output-artifact instructions. The child must not modify project/source files, but returning findings through its normal response or configured output artifact is allowed unless the parent explicitly set `output: false`.\n\nUse `contact_supervisor` with `reason: \"progress_update\"` when:\n- a child is explicitly asked for progress\n- a meaningful discovery changes the plan\n- a long-running child needs to report a blocked/progress checkpoint without waiting for normal tool return flow\n\nMessage conventions:\n- `reason: \"need_decision\"` and `reason: \"interview_request\"` wait for the parent reply and return it to the child.\n- `reason: \"progress_update\"` is non-blocking and should stay concise.\n- Child-side routine completion handoffs are not expected. Native supervisor messages are for decisions, structured input, and meaningful progress updates while a child is still running.\n\nIf bridge instructions provide the child-facing tool, a child can ask:\n\n```typescript\ncontact_supervisor({\n reason: \"need_decision\",\n message: \"Should I optimize for readability or performance here?\"\n})\n```\n\nThe parent replies with the native supervisor tool:\n\n```typescript\nsubagent_supervisor({ action: \"reply\", message: \"Optimize for readability.\" })\n```\n\nOr inspects unresolved asks first:\n\n```typescript\nsubagent_supervisor({ action: \"pending\" })\n```\n\nIf no external `pi-intercom` tool owns the `intercom` name, native supervisor coordination may also expose `intercom` as a compatibility fallback. Prefer `subagent_supervisor` for parent replies because it never overrides installed `pi-intercom`.\n\nIf intercom messages do not show up, run `subagent({ action: \"doctor\" })` or `/subagents-doctor`.\n\n## Management Mode\n\nThe `subagent(...)` tool also supports management actions.\n\n### List available agents and chains\n\n```typescript\nsubagent({ action: \"list\" })\n```\n\n### Create an agent\n\n```typescript\nsubagent({\n action: \"create\",\n config: {\n name: \"my-agent\",\n package: \"code-analysis\",\n description: \"Project-specific implementation helper\",\n systemPrompt: \"Your system prompt here.\",\n systemPromptMode: \"replace\",\n model: \"openai-codex/gpt-5.4\",\n tools: \"read,grep,find,ls,bash\"\n }\n})\n```\n\n### Update an agent\n\n```typescript\nsubagent({\n action: \"update\",\n agent: \"code-analysis.my-agent\",\n config: {\n thinking: \"high\"\n }\n})\n```\n\n### Delete an agent\n\n```typescript\nsubagent({ action: \"delete\", agent: \"code-analysis.my-agent\" })\n```\n\n### Eject, disable, enable, and reset\n\n```typescript\n// Copy a bundled builtin/package agent to user scope as an editable custom file.\nsubagent({ action: \"eject\", agent: \"reviewer\" })\nsubagent({ action: \"eject\", agent: \"reviewer\", agentScope: \"project\" })\n\n// Hide an agent from runtime discovery without deleting it (reversible).\nsubagent({ action: \"disable\", agent: \"reviewer\" })\nsubagent({ action: \"enable\", agent: \"reviewer\", agentScope: \"project\" })\n\n// Delete the scope's custom agent file and/or settings override, restoring the bundled default.\nsubagent({ action: \"reset\", agent: \"reviewer\" })\n```\n\n`eject` copies a builtin or package agent verbatim into the user (default) or project agent dir so it can be customized without hunting package files; the copy shadows the original by runtime name. `disable` writes a reversible `agentOverrides.<name>.disabled: true` entry to the user or project settings file. `enable` removes that `disabled` field while keeping any other override fields. `reset` removes the scope's custom file and settings override to restore the bundled default, and refuses if no bundled default exists (use `delete` for purely custom agents). All four take optional `agentScope: \"user\" | \"project\"`; project overrides win over user ones, so target the project scope to undo a project-scope disable.\n\nUse management actions when the system needs to create or edit subagents on\ndemand without dropping into raw file editing.\n\nManagement actions create or update user/project agent files. `config.name` is the local frontmatter name; optional `config.package` registers and looks up the runtime name as `{package}.{name}`. Use the dotted runtime name for `get`, `update`, `delete`, slash commands, and chain steps. For small builtin changes such as a model swap, prefer `subagents.agentOverrides` in settings.\n\n## Creating and Editing Agents by File\n\nA minimal agent file looks like this:\n\n```markdown\n---\nname: my-agent\npackage: code-analysis\ndescription: What this agent does\nmodel: openai-codex/gpt-5.4\nthinking: high\ntools: read, grep, find, ls, bash\nsystemPromptMode: replace\ninheritProjectContext: true\ninheritSkills: false\nskills: safe-bash, review-checklist\nskillPath: ./skills, ../shared-skills\n---\n\nYour system prompt here.\n```\n\nThat is only a starting point. Omit `package` for the traditional unqualified runtime name. Common optional fields include:\n- `defaultProgress`\n- `defaultReads`\n- `output`\n- `fallbackModels`\n- `subagentOnlyExtensions`\n- `skills`\n- `skillPath`\n- `memory`\n- `maxSubagentDepth`\n- `acceptance`\n- `acceptanceRole`\n\n`acceptance` is a single-agent launch default. Use a scalar level such as `checked` or an inline/block YAML map such as `{ level: \"none\", reason: \"lightweight lookup\" }`. An explicit tool-call value wins; chain and parallel acceptance remains configured on the task or step. Management create/update accepts the same policy object, and `acceptance: \"\"` clears the frontmatter default (`false` remains the deprecated disabled-policy shorthand).\n\n`acceptanceRole` is `read-only` or `writer` and controls automatic acceptance inference only. Explicit task mutation or no-edit intent wins; otherwise the role replaces agent-name guessing. Omission preserves the current name heuristics. The field does not grant or revoke tools. Management accepts `false` or an empty string to clear it.\n\n`tools` is a strict child allowlist, not an extension loader. For a named extension tool, keep its registered name in `tools` and load its provider through normal Pi discovery, `extensions`, a path-like `tools` entry, or `subagentOnlyExtensions`. For example, pair `tools: read, fixture_search` with `subagentOnlyExtensions: ./tools/fixture-search.ts` when the provider should exist only in that agent's child sessions. The child now fails with the unavailable names and provider-loading guidance instead of silently continuing when a requested tool is absent; internal `structured_output` is allowed automatically when an output schema requires it.\n\n`skillPath` adds invocation-private skill files or discovery directories relative to the agent file; it does not select them, so list the desired names under `skills`. Local matches win, unresolved or unreadable matches use normal discovery, and local candidates never enter the parent/global catalog. Use `memory: { scope: \"project\" | \"user\", path: \"<name>\" }` for opt-in role-specific durable memory under the dedicated `agent-memory/` namespace; it is separate from parent/session project memory.\n\nFor many customizations, builtin overrides in settings are lower-friction than\ncopying a full builtin file.\n\n## Prompt Template Integration\n\nThe package includes prompt shortcuts for common workflows: `/parallel-review`,\n`/review-loop`, `/parallel-research`, `/parallel-context-build`,\n`/parallel-handoff-plan`, `/gather-context-and-clarify`, and\n`/parallel-cleanup`. Use them when the user wants repeatable review,\nreview/fix loops, research, context handoff, implementation handoff,\nclarification, or cleanup-review patterns. `/parallel-review autofix` and\n`/parallel-cleanup autofix` synthesize reviewer feedback and then apply only the\nfixes worth doing now. Parent agents can also apply the same recipes directly\nwith `subagent(...)` when the user describes the workflow in natural language\ninstead of invoking a slash command.\n\nAdditional user prompt templates can delegate into `pi-subagents` through the native `/prompt-workflow` and `/chain-prompts` commands. This is useful when a slash command should always run through a particular agent or with forked context. Prompt frontmatter can set `subagent`, `model`, `skill`, `cwd`, `worktree`, `fresh`, `fork`, or `inheritContext` for the native adapter.\n\n## Extension RPC\n\nOther Pi extensions can call `pi-subagents` through the in-process event bus. The stable v1 channels are `subagents:rpc:v1:ready`, `subagents:rpc:v1:request`, and per-request replies at `subagents:rpc:v1:reply:<requestId>`. Envelopes use `{ version: 1, requestId, method, params }`, and replies use `{ version: 1, requestId, success, data | error }`.\n\nMethods: `ping`, `status`, `spawn`, `interrupt`, and `stop`. `spawn` is async-only and rejects management actions, `async: false`, or `clarify: true`; it reuses the normal executor, so discovery, validation, session attribution, configured spawn caps, child-safety depth, artifacts, and async status are shared with the `subagent` tool. `status` and `interrupt` map to the normal control actions. `stop` targets running async runs through the existing timeout control channel. `pi.events` is process-local, so separate Pi processes and child subagents need lifecycle artifact files or `pi-intercom` instead.\n\n## Important Constraints\n\n- **Forking requires a persisted parent session.** If the current session does not\n have a persisted session file, forked runs fail. Packaged `planner`, `worker`,\n and `oracle` default to forked context, so use `context: \"fresh\"` explicitly\n when that is not available or not wanted.\n- **Forked runs inherit parent history.** They are branched threads, not fresh\n filtered contexts. Use fresh context for adversarial reviewers unless the user explicitly asks for forked context.\n- **Default subagent nesting depth is 2.** Deeper recursive delegation is blocked\n unless configured otherwise.\n- **Attention signals are not lifecycle state.** `needs_attention` means no activity has been observed past the configured threshold. `paused` means the child turn was intentionally interrupted or is awaiting direction; it is not the same as `failed`.\n- **Intercom asks are blocking.** A session can only maintain one pending outbound\n ask wait state at a time.\n- **Keep conversational authority clear.** Advisory subagents should not silently\n become second decision-makers.\n\nRuntime config can change orchestration behavior. `asyncByDefault` and `forceTopLevelAsync` affect whether launches detach; `waitTool` can make direct `subagent_wait()` calls return immediately while headless auto-drain remains active, and its effective value is propagated to child runtimes; `globalConcurrencyLimit` bounds concurrent fanout, while a positive `maxSubagentSpawnsPerSession` optionally caps cumulative launches (`0` or unset is unlimited). Status and doctor report the budget; static work preflights declared capacity; only the settled root interactive parent can use `grant-spawn-budget` after native confirmation, with total grants bounded by the original cap. Compaction does not reset usage or grants; `singleRunOutputBaseDir` and `worktreeBaseDir` route outputs and worktrees; `completionBatch` groups async notifications. Per-run `artifacts: false` disables artifact capture for that launch. Async status and result artifacts are versioned with fields such as `lifecycleArtifactVersion`, `workflowGraph`, `steps`, `results`, `totalTokens`, `totalCost`, `turnCount`, `toolCount`, and nested `children`. Child protocol failures expose a structured `protocolError`; `protocol_output_limit` means a child emitted a JSONL line above the 4 MiB live-parser cap. Prefer these artifacts and `status` views over scraping terminal output.\n\n## Best Practices\n\n### Prefer async orchestration\n\nLaunch every subagent asynchronously by default. Use `async: true` for scouts, researchers, workers, reviewers, validators, oracle checks, one-off delegates, chains, and parallel groups unless you intentionally need a foreground/blocking run. The parent should keep moving: inspect code while scouts run, prepare validation while a worker implements, do a local diff pass while reviewers review, and synthesize or verify while a fix worker applies accepted feedback. Async is the default orchestration posture; foreground runs are the explicit opt-out.\n\n### Use subagent_wait() to block until async runs finish\n\nIn an interactive chat, do not call `subagent_wait()` merely to wait after launching background work; return control to the user and Pi will wake the session on completion. Override that default when the current request is run-to-completion — for example, the user asked you to stay with the task and report results back this turn or a skill must finish in one turn. In a headless run, Pi auto-drains exact current-session work at `agent_end`; call `subagent_wait()` when this turn must receive results before it ends. In either case, `subagent_wait()` blocks the current turn until the next run completes or needs attention, keeps the turn alive for normal notification delivery, then returns.\n\n- `subagent_wait()` — return when the next initially active async run or registered provider item finishes, or a subagent needs attention.\n- `subagent_wait({ all: true })` — block until every async run and provider item active at call time finishes, or a subagent needs attention.\n- `subagent_wait({ id: \"...\" })` — block on one async or remembered detached foreground run (id or prefix). Provider items are not selected through this parameter.\n- `subagent_wait({ timeoutMs })` — cap the block; active work keeps running if it elapses.\n\nProviders are discovered through the versioned `pi-subagents/background-work` registry and must return stable item IDs with exact owning session IDs. Child agents receive no provider automatically: keep `subagent_wait` in the child `tools` allowlist and load provider extensions through `extensions` or `subagentOnlyExtensions`.\n\nFor non-interactive fleet orchestration, `subagent_wait()` can keep N workers in flight: launch N, wait for the next completion, react to the result, launch a replacement if needed, then wait again. Use `subagent_wait({ all: true })` only when you intentionally want to drain the fleet to zero. If the turn ends first, headless `agent_end` auto-drain still waits for exact current-session work. In an interactive session, return to the user instead of holding the turn open just to await completion.\n\nIf config or `PI_SUBAGENT_WAIT_TOOL_ENABLED` disables blocking behavior, direct `subagent_wait` calls return immediately. Headless `agent_end` auto-drain remains active as a lifecycle safeguard and surfaces provider, reconciliation, or timeout failures.\n\n### Keep writes single-threaded by default\n\nA strong pattern is one main decision-maker plus advisory/research/review/validation subagents around it. Use `oracle` for advice and `worker` for the actual write path. Parallelize reading, review, validation, and synthesis support, not normal writes, unless you deliberately isolate writers with worktrees. A child that writes should report what changed, what was left undone, commands run with exit codes, validation evidence, surprises, and any decisions that need parent approval.\n\n### Use fork for branched advisory or execution threads\n\nForked runs are useful when the child should reason in a separate thread while\nstill inheriting the parent’s accumulated context. They are especially useful for\n`oracle`, which audits inherited decisions and drift. For adversarial code review,\nprefer fresh-context reviewers that inspect the repo and diff directly unless the\nuser explicitly requests forked context.\n\n### Prefer narrow tasks\n\nGive subagents specific tasks rather than vague mandates.\n`Review auth.ts for null-check gaps` works better than `Review everything`.\n\n### Escalate decisions upward\n\nIf a subagent encounters an unapproved product, architecture, or scope choice,\nit should use `contact_supervisor` and wait for the reply instead of deciding alone. Generic `intercom` is a fallback only when the bridge-provided supervisor tool is unavailable.\n\n### Intervene only on clear control signals\n\nUse subagent control proactively when a delegated run emits `needs_attention`, or when a human asks you to regain control. Do not interrupt just because a child has briefly produced no output. Silence can be normal during long tool calls, test runs, or model reasoning.\n\n### Name sessions meaningfully\n\nUse `/name` so intercom targeting stays stable.\n\n## Common Workflows\n\n### Recon → Plan → Implement\n\n```typescript\nsubagent({\n chain: [\n { agent: \"scout\", task: \"Map the auth flow and summarize relevant files\" },\n { agent: \"planner\", task: \"Plan the migration from {previous}\" },\n { agent: \"worker\", task: \"Implement the approved plan from {previous}\" }\n ]\n})\n```\n\n### Fable mode for complex work\n\nFable mode is the default orchestration posture for complex work. It is not a separate runtime mode; it is how the parent session uses `subagent`, `interview`, `subagent_wait`, acceptance contracts, artifacts, and fresh-context review when the work has real complexity. Use it for complex features, broad refactors, migrations, ambiguous goals, multi-system changes, expensive validation, user-visible behavior changes, or any request to plan/orchestrate end to end. Do not force it onto tiny one-shot delegation.\n\nRun the work through seven gated phases:\n\n1. **Understand** — use `scout` or `context-builder` fanout for breadth, but the parent personally reads the load-bearing files and lets direct source reading decide disagreements. Gate: the parent can quote the exact code or behavior being changed and knows the repo's verification harness.\n2. **Decide** — separate user-owned decisions from implementation judgments. Use `interview` for product, naming, cost, taste, or risk decisions; decide routine engineering details in the parent and state them. Gate: every user-owned decision needed for design is answered.\n3. **Design** — use `planner`, `context-builder`, or read-only design/review children for parallel perspectives. Before parallel workstreams, write seam contracts: ownership boundaries, composition points, assumptions, and validation handoffs. Gate: one parent-synthesized plan and written seams for parallel work.\n4. **Implement** — capture a baseline first, then launch one async `worker` as the sole writer for the active worktree unless isolated worktrees were intentionally requested. Break large work into serial milestones instead of concurrent writes. Gate: build/typecheck is green and every output or diff delta is characterized as intended or fixed.\n5. **Verify** — climb the spend ladder: static checks, free end-to-end/dry-run, cheapest live probe, targeted changed-path live test, then full realistic run when warranted. Observe the artifact itself, not only exit codes or scores, and confirm the changed code actually executed. Gate: the highest necessary rung has directly observed evidence matching intent.\n6. **Iterate** — when a gate or reviewer finds a defect, the parent names the failure class, searches for siblings, synthesizes fixes, and sends exactly one fix worker for accepted changes. For LLM judges, gates, or detectors, trigger on concrete findings rather than scores, record pass/violations/error verdicts, cache nondeterministic verdicts by input hash, budget enough output tokens, and sanitize judge text before reusing it downstream. Gate: the class is fixed or explicitly bounded, and recurrence detection exists when feasible.\n7. **Ship** — run adversarial fresh-context review/validation outside the implementation path, disposition every finding, rerun affected gates, then have the parent inspect the final diff. Commit, push, release, or open PRs only inside user-approved boundaries. Gate: findings are dispositioned, gates re-pass, and the final summary names evidence, artifacts, residual risks, and output paths.\n\n### Clarify → Plan → Implement → Review (self-orchestrated workflow)\n\nFor straightforward non-trivial work, this sequence is the lightweight version of the parent-owned loop. When the task is complex, use Fable mode above. In either case, factor in the packaged prompt workflows without literally invoking slash commands. Use the same patterns through tools and subagents.\n\nKeep builtin agent defaults unless the user explicitly asks for a different model, thinking level, skills, output behavior, context mode, or other override. Do not add overrides just because you are orchestrating; the defaults encode the intended role behavior. In particular, packaged `planner`, `worker`, and `oracle` default to forked context.\n\nWhen the user approves launching a subagent to carry out a plan or workflow, treat that as approval to generate a proper role-specific meta prompt for that subagent. Include the approved plan path or summary, clarified requirements, non-goals, relevant context, role boundaries, files or areas to inspect, acceptance criteria, expected output, and validation expectations. Do not pass vague instructions like “implement the plan fully” or “review this” by themselves.\n\n- `/gather-context-and-clarify` maps to: launch `scout` and, when needed, `researcher`; synthesize findings; then use `interview` to ask every clarification question needed for shared understanding.\n- `/parallel-review` maps to: launch fresh-context `reviewer` agents with distinct review angles; synthesize the feedback before applying anything.\n- `/review-loop` maps to: keep the parent in charge of worker → fresh reviewers → synthesized fix worker cycles until no fixes worth doing now remain, an unapproved decision appears, or the review-round cap is reached.\n- `/parallel-research` maps to: combine local `scout` context with external `researcher` evidence when current docs, ecosystem behavior, or API details matter.\n- `/parallel-context-build` maps to: run a chain-mode parallel group of `context-builder` agents with distinct temp output paths, then synthesize their context and meta-prompt sections.\n- `/parallel-handoff-plan` maps to: run external `researcher` plus local/strategy `context-builder` passes, then a synthesis `context-builder` that writes an implementation handoff plan and implementation-ready meta-prompt.\n- `/parallel-cleanup` maps to: use review-only cleanup passes after implementation, especially for simplicity, verbosity, and redundant tests.\n\nFor feature work, use this sequence as scaffolding for parent-agent behavior:\n\n```text\nclarify → validation contract → planner → async worker → parallel async fresh-context reviewers/validators → async fix worker → follow-up review when warranted → parent review\n```\n\nThe validation contract defines acceptance before code is written: expected behavior, acceptance checks, commands or user flows to exercise, and evidence the worker should return. Keep it lightweight for small tasks, but make it explicit enough that reviewers and validators are checking the intended outcome rather than the worker’s own assumptions.\n\nUse the structured `acceptance` field when the run should carry an explicit acceptance contract. If omitted, subagents infer an effective acceptance policy from role, mode, and risk. Use `level: \"checked\"` for ordinary writer evidence gates and `level: \"verified\"` when the runtime should run explicit validation commands. Do not explicitly request `level: \"reviewed\"`: the current run cannot supply an independent reviewer result, so that level is reserved for inferred policy. Orchestrate a separate reviewer instead. To disable gates, use `{ level: \"none\", reason: \"...\" }`; the bare string `\"none\"` is rejected, and `false` is accepted only as a deprecated shorthand. Do not call a run reviewed just because the worker says it is done; reviewed means a reviewer gate returned a result. Child-reported command success is evidence, not runtime verification.\n\nThe first `worker` implements the approved plan. The parent continues with independent inspection or validation prep while it runs, not parallel edits to the same worktree. When the async worker completes, treat its handoff as the transition into review, not as final completion, unless the user explicitly asked for worker-only work, review-only output, or to stop after implementation. Parallel reviewers inspect the resulting diff from fresh context. Validators check behavior with the best available evidence: commands, tests, browser/CLI interaction, screenshots, logs, or manual reproduction notes. The final `worker` applies synthesized review fixes in forked context, then the parent looks over the final diff before completing. The parent may launch these steps as an initial async chain when the workflow is already clear, or as follow-up subagent runs after each async completion. Initial chains should pass `async: true` so the main chat is unblocked; avoid `clarify: true` unless the user asked for foreground clarification. Do not stop after parallel review unless the user explicitly asked for review-only output or the review surfaced a decision that needs approval first.\n\nFor complex work, risky changes, broad refactors, or many changed lines, increase review and validation fanout rather than trusting one reviewer. Use distinct angles such as correctness/regressions, tests/validation, simplicity/maintainability, security/privacy, performance, docs/API contracts, and user-flow behavior. When reviewers find non-trivial issues or the fix worker touches many lines, run another focused review round before final validation.\n\nWhen review has already produced concrete findings across several independent areas, use staged fix orchestration: parallel read-only planners for each issue cluster, one sole writer worker for the active worktree, then parallel fresh-context validators. This is the safest way to handle a dirty worktree with many prior changes because it parallelizes judgment without parallelizing writes. Non-blocking suggestions may go into the writer prompt only if they are small, safe, and inside the approved scope; otherwise defer them explicitly.\n\nFor very large work, split into serial milestones instead of launching a swarm of writers. Each milestone gets one writer, a validation contract, fresh-context review/validation, a fix pass, and parent acceptance before the next milestone starts. Use parallel subagents inside a milestone for read-only context, research, review, and validation only.\n\nKeep orchestration authority in the parent session. Child subagents should not launch more subagents, read this skill, or run their own orchestration loops unless the parent intentionally selected a fanout agent whose builtin `tools` includes `subagent`. Spawned subagents do not receive the `pi-subagents` skill, parent-only status/control/slash messages, or prior parent `subagent` tool-call/tool-result artifacts. Ordinary children also do not receive the `subagent` extension tool. Child context filtering strips old hidden orchestration-instruction messages when they appear in inherited history. Every child receives a boundary instruction: ordinary children are told the parent owns orchestration and they must not propose or run subagents; explicit fanout children are told to use `subagent` only for the assigned fanout work, with `maxSubagentDepth` still enforced. Implementation children must call real edit/write tools instead of printing pseudo tool calls. Pass children concrete role-specific work instead.\n\n1. Clarify first. This is mandatory. Gather code context with `scout` or `context-builder`, add `researcher` only when external evidence matters, then ask the user clarifying questions with `interview` until scope, acceptance criteria, constraints, and non-goals are clear.\n2. Define the validation contract. State acceptance before implementation: expected behavior, checks to run, user flows to exercise, and evidence required in the worker handoff. For UI, CLI, integration, or workflow changes, include at least one validator angle that uses the product the way a user would rather than only reading code.\n3. Plan when useful. For complex work, call `planner` or write a plan doc yourself and get approval before implementation. For simple work, confirm shared understanding and explicitly note why planning is skipped.\n4. Implement with one writer. After approval, launch `worker` asynchronously with a proper meta prompt that includes clarified requirements, relevant context, plan path or summary, the validation contract, and output expectations. Packaged `worker` defaults to forked context; pass `context: \"fresh\"` only when you intentionally want a fresh child. While it runs, prepare validation or inspect adjacent code instead of editing the same worktree.\n5. Require a useful worker handoff. Ask the worker to report changed files, what was implemented, what was left undone, commands run with exit codes, validation evidence, surprises or new risks, decisions made inside approved scope, and decisions needing parent approval.\n6. Review after implementation. After the worker completes, launch parallel async fresh-context `reviewer` agents for correctness/regressions, tests/validation, and simplicity/maintainability. Add security, performance, docs/API, domain-specific, or user-flow validators for complex work, risky changes, broad refactors, or many changed lines. Use `output: false` unless review artifacts are explicitly needed.\n7. Synthesize, then run the fix worker. Separate blockers, fixes worth doing now, optional improvements, and feedback to ignore/defer, then launch an async forked `worker` to apply fixes worth doing now when the workflow is implementation-authorized. If reviewers found scope/product/architecture choices that were not approved, ask the user first instead of applying them.\n8. Review again when warranted. If the fix worker made substantial changes or addressed non-trivial findings, run another focused parallel review round before final validation.\n9. Validate and complete. After the fix worker and any follow-up review return, inspect the final diff yourself, run or confirm focused validation, update docs/changelog when relevant, and summarize what changed and why.\n\nExample implementation handoff after clarification and optional planning:\n\n```typescript\nsubagent({\n agent: \"worker\",\n task: \"Implement the approved feature.\\n\\nClarified requirements:\\n- ...\\n\\nPlan: see ~/Documents/docs/...-plan.md\\n\\nValidation contract:\\n- ...\\n\\nReturn a handoff with changed files, what was implemented, what was left undone, commands run with exit codes, validation evidence, surprises/new risks, and decisions needing parent approval.\",\n acceptance: {\n level: \"checked\",\n evidence: [\"changed-files\", \"tests-added\", \"commands-run\", \"residual-risks\", \"no-staged-files\"]\n },\n async: true\n})\n```\n\nExample review pass after implementation:\n\n```typescript\nsubagent({\n tasks: [\n { agent: \"reviewer\", task: \"Review the current diff for correctness and regressions. Inspect changed files directly; do not rely on the worker's reasoning.\", output: false },\n { agent: \"reviewer\", task: \"Review the current diff for tests and validation quality against the validation contract. Inspect changed files directly.\", output: false },\n { agent: \"reviewer\", task: \"Review the current diff for simplicity and maintainability. Inspect changed files directly.\", output: false }\n ],\n concurrency: 3,\n context: \"fresh\",\n async: true\n})\n```\n\nExample fix worker after parallel reviews:\n\n```typescript\nsubagent({\n agent: \"worker\",\n task: \"Apply the synthesized reviewer feedback below. Only apply fixes worth doing now; preserve user-approved scope; ask before unapproved product or architecture changes. Run focused validation and summarize what changed.\\n\\nReviewer synthesis:\\n...\",\n async: true\n})\n```\n\n### Review loop\n\nDo not treat review as the final step for implementation work. Run reviewers and validators, synthesize their findings against user scope and the validation contract, then launch one `worker` for accepted fixes when implementation is authorized.\n\nWhen an async implementation worker completes, treat the worker handoff as an intermediate state. The next parent action is review fanout, then synthesis, then a fix worker if reviewers found fixes worth doing now. This can be planned as an initial async chain when the whole workflow is known, or continued as follow-up subagent runs when the parent only launched the first worker initially. Initial chains should pass `async: true` so the main chat is unblocked; `clarify: true` is the explicit foreground opt-in.\n\nFor explicit review-loop requests, repeat worker → fresh-reviewer → synthesized-fix-worker cycles until reviewers find no blockers or fixes worth doing now, remaining feedback is optional or intentionally deferred, an unapproved product/scope/architecture decision needs the user, or the max review-round cap is reached. Default to 3 review rounds unless the user sets a different cap. For complex work, many changed lines, or any fix pass that materially changes the diff, run another focused review round before the parent’s final look; otherwise stop instead of chasing optional polish.\n\n### Parallel non-conflicting analysis\n\n```typescript\nsubagent({\n tasks: [\n { agent: \"scout\", task: \"Audit frontend auth flow\" },\n { agent: \"researcher\", task: \"Research current retry/backoff best practices\" }\n ]\n})\n```\n\n### Saved chain\n\n```text\n/run-chain review-chain -- review this branch\n```\n\nUse saved `.chain.md` or `.chain.json` workflows when the user wants a repeatable multi-agent flow without rewriting the chain each time. Prefer `.chain.json` for dynamic fanout or inline `outputSchema` objects; `.chain.md` remains the simple sequential/static authoring format.\n\n## Error Handling\n\n**\"Unknown agent\"**\n```typescript\nsubagent({ action: \"list\" })\n// Check available agents and chains, then confirm scope/precedence.\n```\n\n**Setup, discovery, or intercom confusion**\n```typescript\nsubagent({ action: \"doctor\" })\n// Check runtime paths, async support, discovery counts, current session, and intercom bridge state.\n```\n\n**\"Max subagent depth exceeded\"**\n```typescript\n// Flatten the workflow or raise maxSubagentDepth in config.\n```\n\n**\"Session manager did not return a session file\"**\n```typescript\n// Persist the current session before using context: \"fork\".\n```\n\n**Intercom \"Already waiting for a reply\"**\n```typescript\n// Resolve the current outbound ask before starting another one.\n```\n\n**Parallel output-path conflict**\n```typescript\n// Give each parallel task a distinct output path, or disable output for tasks that do not need it.\n```\n\n**Worktree launch fails**\n```typescript\n// Ensure the git working tree is clean and task cwd overrides match the shared cwd.\n```\n\n**Child fails before starting**\n```typescript\n// Inspect `subagent({ action: \"status\", id: \"...\" })`, artifact metadata/output logs, and run doctor. Extension loader errors usually appear in child output logs.\n```\n</skill>"}],"timestamp":1784878804386}}
|
||
{"type":"message","id":"0a8fd473","parentId":"0429a573","timestamp":"2026-07-24T07:40:07.286Z","message":{"role":"assistant","content":[{"type":"thinking","thinking":"Got it. The user has activated context-mode with a hierarchy and is in investigate mode. I need to acknowledge this and be ready to use context-mode tools. The current directory is /home/liph/dotfiles/pi/.pi/agent. No specific task yet beyond acknowledging the context. I should confirm understanding briefly and ask what they want to investigate or work on.","thinkingSignature":"reasoning"},{"type":"text","text":"Context-mode hierarchy acknowledged. I’m in **investigate** mode and will route data-heavy work through the context tools:\n\n- Multi-command / research → `ctx_batch_execute`\n- Derive-from-data → `ctx_execute` / `ctx_execute_file`\n- Web docs → `ctx_fetch_and_index` then `ctx_search`\n- Local docs → `ctx_index`\n- Diagnostics → `ctx_doctor`, stats → `ctx_stats`, purge → `ctx_purge`\n\nWhat do you want to investigate?"}],"api":"openai-completions","provider":"ollama","model":"kimi-k2.7-code:cloud","usage":{"input":31994,"output":177,"cacheRead":0,"cacheWrite":0,"reasoning":0,"totalTokens":32171,"cost":{"input":0,"output":0,"cacheRead":0,"cacheWrite":0,"total":0}},"stopReason":"stop","timestamp":1784878804755,"responseId":"chatcmpl-914","responseModel":"kimi-k2.7-code"}}
|
||
{"type":"message","id":"e9a146b4","parentId":"0a8fd473","timestamp":"2026-07-24T07:40:46.999Z","message":{"role":"user","content":[{"type":"text","text":"nothing it was just a test"}],"timestamp":1784878846902}}
|
||
{"type":"message","id":"627e12ae","parentId":"e9a146b4","timestamp":"2026-07-24T07:40:49.545Z","message":{"role":"assistant","content":[{"type":"thinking","thinking":"Context-mode hierarchy acknowledged. I’m in implement mode and will route data-heavy work through the context tools:\n\n- Multi-command / research → `ctx_batch_execute`\n- Derive-from-data → `ctx_execute` / `ctx_execute_file`\n- Web docs → `ctx_fetch_and_index` then `ctx_search`\n- Local docs → `ctx_index`\n- Diagnostics → `ctx_doctor`, stats → `ctx_stats`, purge → `ctx_purge`\n\nGot it — just a test. I’m ready when you have a real task.","thinkingSignature":"reasoning"}],"api":"openai-completions","provider":"ollama","model":"kimi-k2.7-code:cloud","usage":{"input":32111,"output":110,"cacheRead":0,"cacheWrite":0,"reasoning":0,"totalTokens":32221,"cost":{"input":0,"output":0,"cacheRead":0,"cacheWrite":0,"total":0}},"stopReason":"stop","timestamp":1784878847000,"responseId":"chatcmpl-44","responseModel":"kimi-k2.7-code"}}
|
||
{"type":"message","id":"32ce9bed","parentId":"627e12ae","timestamp":"2026-07-24T07:48:55.104Z","message":{"role":"user","content":[{"type":"text","text":"/skill:user-install"}],"timestamp":1784879335022}}
|
||
{"type":"message","id":"21dc9ae9","parentId":"32ce9bed","timestamp":"2026-07-24T07:49:01.776Z","message":{"role":"assistant","content":[{"type":"thinking","thinking":"We need respond to user. They said nothing it was just a test, then slash command /skill:user-install. The system likely expects to handle slash command. The user invoked skill user-install. We should read the skill file? The skill is for babysitter user-install; but this is a test. We need perhaps acknowledge. We have session_mode implement. We need see if skill needs loaded. The slash command means they want to run user-install skill? It says use read tool to load skill when task matches. Should we read skill file and follow? Since user said nothing it was just a test, maybe they don't actually want to install. But they invoked the slash. We could respond that it was a test and offer to proceed? The instruction: \"Use the read tool to load a skill's file when the task matches its description.\" So when they invoke /skill:user-install, we should read the skill file. But they said it was just a test. Possibly they tested slash command and want a response. We should maybe read the skill file to show it's recognized? But if it's a test, we can reply that context-mode is active and user-install skill recognized, ask if they'd like to proceed. We shouldn't launch an install without confirmation. Also session state says implement mode. Maybe answer concisely: acknowledged, no action taken because it was a test; if you want user-install, confirm. We may also mention babysitter user-install details. But there is a skill instruction to read file when task matches. Should we read? Could just state: \"Got it — that was a test. `/skill:user-install` would run the Babysitter onboarding flow, but I won't start it without your confirmation. Let me know if you want to proceed.\" That's safe. Let's include maybe path. Final concise.","thinkingSignature":"reasoning"},{"type":"text","text":"Got it — that was just a test.\n\n`/skill:user-install` would launch the **Babysitter user-install** onboarding flow (builds your user profile, installs dependencies, configures workflow tools). I won’t run it unless you confirm, since onboarding is user-specific.\n\nIf you actually want to run it, just say “yes, run user-install.”"}],"api":"openai-completions","provider":"ollama","model":"kimi-k2.7-code:cloud","usage":{"input":32119,"output":441,"cacheRead":0,"cacheWrite":0,"reasoning":0,"totalTokens":32560,"cost":{"input":0,"output":0,"cacheRead":0,"cacheWrite":0,"total":0}},"stopReason":"stop","timestamp":1784879335105,"responseId":"chatcmpl-545","responseModel":"kimi-k2.7-code"}}
|