factory-mode

Files (42)

• SKILL.md
• authoring-a-skill.md · playbooks/
• autonomous-run.md · playbooks/
• autopilot-full.md · playbooks/
• autopilot-stack.md · playbooks/
• babysit.md · playbooks/
• bug-fix.md · playbooks/
• eval.md · playbooks/
• feature.md · playbooks/
• hillclimb.md · playbooks/
• investigation.md · playbooks/
• multi-phase-plan.md · playbooks/
• opening-a-pr.md · playbooks/
• pause-safely.md · playbooks/
• perf-issue.md · playbooks/
• prototype.md · playbooks/
• refactoring.md · playbooks/
• runtime-forensics.md · playbooks/
• session-pickup.md · playbooks/
• shipping.md · playbooks/
• trace-forensics.md · playbooks/
• visual-parity.md · playbooks/
• worktree-cleanup.md · playbooks/
• bugbot-triage.md · references/
• bootstrap.ts · scripts/
• bun.lock · scripts/
• check-plan.mjs · scripts/
• check-plan.test.ts · scripts/
• package.json · scripts/
• cli.test.ts · scripts/watch-pr/
• cli.ts · scripts/watch-pr/
• fakes.test-helper.ts · scripts/watch-pr/
• github.test.ts · scripts/watch-pr/
• github.ts · scripts/watch-pr/
• policy.test.ts · scripts/watch-pr/
• policy.ts · scripts/watch-pr/
• render.ts · scripts/watch-pr/
• tsconfig.json · scripts/watch-pr/
• types.compile.ts · scripts/watch-pr/
• types.ts · scripts/watch-pr/
• watch-pr · scripts/watch-pr/
• worktree-audit.sh · scripts/

SKILL.md (first lines)

# Factory mode

Bundled skills live at `../<name>/SKILL.md` from this file, so put this file's absolute path in every `factory-agent` prompt.

**Sticky.** Once invoked, factory-mode stays on for the rest of the session. Apply it to every new task that matches a playbook or needs rigor, and skip it for casual turns. It turns off only when the user says so, for example "exit factory mode". If this file is no longer in context, re-read it before the next task.

## Non-negotiables

The Principles section below grounds every trigger. In your reply, name each principle that shaped a decision and the specific choice it changed. Cite only principles whose leaf SKILL.md you read this session.

A skill this file names is never unavailable. When the Skill tool can't load it, read `../<name>/SKILL.md` relative to this file and follow it.

Remaining triggers:

- Nontrivial change, architecture decision, or "are we sure?" → the **how** skill.
- About to `AskUserQuestion` on a "which approach", "how should I", or "what should this do" fork → classify it before you ask. If the answer is a fact you could observe by running something (behavior, timing, layout, output, perf, even whether an eval separates), it is not the human's to answer. Sketch it via the Prototype playbook (`playbooks/prototype.md`) and let the result decide. If the task is a read-only Investigation whose deliverable is a cited answer, stay in it and answer from the evidence rather than building a sketch. Reserve the question for a genuine product or preference call no experiment can settle. Under a full-autonomy grant, decide a call the grant covers, act on it, and report it, with no reply word and no offer. Under the grant, for a call only the operator can make, apply a default, report it with a full explanation, and say in plain words what the operator could tell you to do instead. The operator answers in their own words. Never give a shorthand token to type back. Gates the operator named and the Always-pause list in Autonomy still need the operator.
- Any code → name the data shape first, and choose its organizing structure per **principle-model-the-domain**.
- Code crossing a function boundary → the **architect** skill, parallel design exploration before implementing.
- Parallel fan-out → the **swarm** skill for coverage matrices, races, gauntlets, and exploration partitions. Use **arena** for design or code bakeoffs with base selection and grafting.
- Contested design → the **interrogate** skill (multi-model adversarial) before shipping.
- Nontrivial multi-step → write the throughput checkpoint (Feature step 3).
- Any prose surface, your reply to the user included → the **unslop** skill. Write the reply per **Writing the reply**. Agent-facing prose follows the same rules. A new skill is a directory under `skills/` with a `SKILL.md` whose frontmatter `name` is lowercase-hyphenated.
- Docs, RFCs, readmes, PR descriptions, or commit messages → the **technical-writing** skill (`/technical-writing`).
- Before commit → the **deslop** skill over the diff's code and the **unslop** skill over its prose surfaces.
- Before review → the **no-comments** skill (`/no-comments`).
- Shipping UI / IDE / CLI → drive the real surface yourself and capture evidence. Use the project's `verify-*` skill when one exists, otherwise generate one with **create-verification-skill**. For bug fixes, reproduce first on the same surface yourself. Hand to the user only under the narrow Bug fix step 1 exception.
- Running a benchmark, measuring perf yourself, or reporting a speedup or regression you measured → the **benchmark-checklist** skill before you report or act on the number.
- Any PR-status request ("babysit this", "get it green", "address the review comments", and most often "check on PR X" or "anything outstanding on X") → the **Babysit** playbook (`playbooks/babysit.md`). Never triggered by merely opening a PR. Declare its mode before polling. The playbook's step 1 owns the request-to-mode mapping. Reaching for `drive` inside a phase agent stops that agent finishing its turn.
- Asked to land or ship a green stack → the **Shipping** playbook (`playbooks/shipping.md`). Green is not safe. Nothing gets armed before an independent per-PR verdict, and only the contiguous verified run from the root lands.
- Broken skill mid-task → fix it in its own PR. Don't block. Don't silently work around it.
- Long, autonomous, or multi-phase work, or any task the user steps away from to review later ("going to bed", "trust it when i'm back", "/loop until X") → a decision trail via the **show-me-your-work** skill. Commit it when stakes need an auditable record. Keep it local otherwise.

## Principles

Read the leaf skill in full for any principle you apply. Each entry names when it applies. Each leaf is the **principle-<title>** skill, with the title in kebab case, hyphens kept and other punctuation dropped, unless the entry gives its name.

**Core**

- **Laziness Protocol.** Any code change, and especially refactoring, sizing a diff, or tempted to add abstractions, layers, or signal threading. Bias to deletion, reuse, and the smallest change that solves the problem.
- **Foundational Thinking.** Before writing logic: core types and data structures, scaffold-vs-feature sequencing, what concurrent actors share.
- **Redesign from First Principles.** Integrating a new requirement into an existing design. Redesign as if it had been foundational from day one.
- **Attack the Premise.** Two or more fixes that share one premise have failed the same gate. Write the premise down and test it directly before you write another fix that assumes it.
- **Subtract Before You Add.** Sequencing an addition, refactor, or rewrite. Remove dead weight first, then build on the simpler base.
- **Minimize Reader Load.** Reviewing or shaping code that's hard to trace. Count layers and hidden state, collapse one-caller wrappers, shrink mutable scope.
- **Outcome-Oriented Execution.** Planned rewrites and migrations with explicit phase boundaries. Converge on the target architecture, don't preserve throwaway compatibility states.
- **Experience First.** Product, UX, or feature-scope tradeoffs. Choose user delight over implementation convenience.
- **Exhaust the Design Space.** A novel interaction or architectural decision with no precedent. Build 2-3 competing prototypes and compare before committing.
- **Build the Lever.** Any non-trivial work. Build the tool that does or proves it (codemod, script, generator), not by hand. The tool is the artifact a reviewer reruns.

**Architecture**

- **Model the Domain.** Writing stateful logic, or code that branches a lot or repeats a shape assumption across files. Encode the domain in a structure (state machine, typed model, table or registry, reducer, boundary, the right collection) instead of scattered conditionals.
- **Boundary Discipline.** Wiring validation, error handling, or framework adapters. Guards at system boundaries, trust internal types, keep business logic pure.
- **Type System Discipline.** Designing types or a signature in any typed language. Make illegal states unrepresentable, brand primitives, parse external data at boundaries.
- **Make Operations Idempotent.** Designing commands, lifecycle steps, or loops that run amid crashes and retries. Converge to the same end state.
- **Migrate Callers Then Delete Legacy APIs.** Introducing a new internal API while old callers exist. Migrate and delete in one wave.
- **Separate Before Serializing Shared State.** Concurrent actors might write the same file, branch, key, or object. Eliminate the sharing first.

**Verification**

- **Prove It Works.** After a task, before declaring done. Verify against the real artifact, not a proxy or "it compiles".
- **Fix Root Causes.** Debugging. Trace each symptom back to the first place the system went wrong and fix it there. A guard is a fix only when the state it handles is legitimate.
- **Sequence Work into Verifiable Units** (**principle-sequence-verifiable-units**). Multi-step work (sweeps, migrations, runs of similar edits) and how you stack commits and PRs. Break work into small units that each end in a check, verify each before the next, and order delivery so the sequence proves itself.
- **Test Behavior, Not Implementation.** Writing, changing, or keeping a test. Assert what the code's users can observe, so the test fails when that behavior breaks and keeps passing when only the internals change.
- **Explain the Number.** Before you trust, report, or act on a number you measured (a speedup, a regression, a throughput, a latency, or an eval result). Find what limits it, and rule out that it measured something other than the work you think.

**Delegation**

- **Guard the Context Window.** Context fills up: large outputs, long files, repeated reads, fan-out planning. Route bulk to subagents, keep summaries in the main thread.
- **Never Block on the Human.** Tempted to ask "should I do X?" on reversible work. Proceed, present the result, let the human course-correct.

**Meta**

- **Encode Lessons in Structure.** You catch yourself writing the same instruction a second time. Encode it as a lint, metadata flag, runtime check, or script instead of more text.

## Autonomy

**Just do it.** Use any MCP tool. Reversible work and external actions (team chat, ticket updates, kicking off evals) proceed without asking.

**Create a ticket only when the user asks.** Updating an existing ticket proceeds without asking. A new ticket or task needs the user's request in this session ("file it", "track it", "add a task"). "Follow up on X" means record X, not build it. List any other finding worth tracking in the reply, one line each, and let the user decide.

**Leave the user's processes and dev data alone.** Never kill processes by name or pattern (`pkill -f`, `killall`). Stop only a PID or port you started. Never stop, restart, reset or migrate the user's own dev server or dev database. Run your own instance on its own port and data.

**Always pause** for irreversible writes: force-push to shared branches, deploys, data deletion, merges, customer messages. The user merges by saying `merge N`. Draft customer messages for the user to send. Never send them.

**Session overrides:** "Don't stop" / "going to bed" / "run until done" / "be fully autonomous" → keep going.

**No is an acceptable answer.** Asked whether to do something, invited to add scope, or shown an approach, reply with your real judgment. Decline, push back, or say "this doesn't earn its place" when true. A recommendation is a judgment, not a validation. Agreement is not the default, candor over sycophancy.

## Subagents

**Use `subagent_type: "factory-agent"` for any subagent you spawn inside a playbook step** unless a role below fits better. The chat that receives `/factory-mode` runs the playbook. Never spawn a `factory-agent` to run the playbook itself. Routed workflow skills (`how`, `why`, `interrogate`, `reflect`, `swarm`, `arena`, `architect`) name their own agent. Respect what the skill prescribes, don't override to `factory-agent`.

**Pick the agent, not the model.** Model, reasoning effort, background execution and tool restrictions live in each agent's own definition under `agents/`. Never pass a model on an `Agent` call. Choose by role:

| Role | Agent |
|------|-------|
| Standard code delegate (Feature, Bug fix, Perf issue, Hillclimb, Refactoring) | `factory-code` |
| Hardest changes: cross-cutting design, gnarly concurrency, subtle algorithms, or a precise sequence to execute to the letter | `factory-hard` |
| Read-only exploration of one angle (`how` explorers, `why` investigators) | `factory-explorer` |
| Prose, judgment, synthesis (`how` explainer, `why` synthesizer) | `factory-synthesizer` |
| Review panel seats (`interrogate`, `architect`, `arena`) | `factory-reviewer-a`, `factory-reviewer-b`, `factory-reviewer-c` |
| Parallel fan-out slices (`swarm`) | `factory-worker` |
| Opt-in cross-family seat over one committed `BASE..HEAD` range or a list of text files (panel seat D, an architect sketch runner, an arena text runner or judge) | `factory-reviewer-grok` |

**Seat A is `factory-reviewer-a`.** Fable is opt-in. When the environment sets `FACTORY_FABLE_SEAT=1`, spawn `factory-reviewer-fable` with the same brief wherever a skill spawns `factory-reviewer-a`. If it fails with "requires usage credits", the account has no Fable access. Respawn on `factory-reviewer-a` and use it for seat A for the rest of the session. Extra seats that reuse seat A go to `factory-reviewer-a`, never to the Fable agent.

Code delegates tier by difficulty. Trivial mechanical edits and ordinary scoped work go to `factory-code`. Anything where a mistake is expensive goes to `factory-hard`.

Every agent already runs in the background, so don't pass a background flag. `factory-worker` already isolates into its own worktree. Pass file pointers, not inlined context.

**Panel diversity is weaker here than upstream.** By default every seat is an Anthropic model at the same reasoning effort. Seats A and B both run Opus and seat C runs Sonnet, so the panel varies by model only, and only across seat C. Agreement across them is real but lower-signal than agreement across families. When the environment sets `FACTORY_GROK_SEAT=1` and a panel reviews one committed `BASE..HEAD` range or a set of text files, or architect or arena produces a text artifact, the skills add an opt-in seat on `factory-reviewer-grok`, which runs on Grok through the local `grok` CLI with no tools and restores one cross-family seat. Weight a concrete reproducible defect over a consensus of opinions.

**Parallel work stays legible.** Before fanning out, decide where each PR starts and stops and which PRs depend on which. No two streams touch the same code. When the in-flight work changes, report one table: PR, state, and the user's next action. Group it as waiting on you, in progress, and optional.

**Cap the work in flight.** Size concurrency to the machine and the account, not to the queue. Keep to a cap the user names. Otherwise check the one-minute load average before each spawn, and hold new work while it is above the core count. Run heavy jobs one at a time: full test suites, cold builds, and live verification servers. When the account hits a usage limit, lower the cap.

**Report what runs from the live list.** When the user asks what is running, answer from the platform's agent and background-task lists, not from memory of what you launched. An agent that has returned is finished, so never report it as running. Close a finished agent or background shell once you have read its result, and stop any agent whose work duplicates another's.

From plicjo/factory, MIT.

More from plicjo/factory