factory-skills

Generate a project-local verification skill that drives the real app and captures evidence for factory lifecycle gates. Use when a project has no scripted way to prove UI, CLI, or service behavior.

Research and plan one ticket, producing a local plan, dependency-ordered task files, and only the durable context or ADR updates the ticket earns. Use before implementation when a feature, bug, refactor, performance task, or investigation needs to become executable work.

Prove one task's acceptance criteria with observable end-to-end tests against the real product, record a test plan and a test-gap ledger, and capture screenshots for the pull request. Use for the QA stage, after work and before review, when FACTORY.json enables stages.qa.

Mine completed ticket evidence for reusable factory improvements, propose narrowly routed skill or tooling changes, and compare candidates against a budget-aware evaluation set before human approval. Use after meaningful factory experience, not after every routine ticket.

Adversarially review one ready-for-review ticket task in a read-only Pi subagent using a model different from the worker, then return a structured markdown findings list and a verdict. Use only for the review stage; it never edits code or applies its own findings.

Implement one dependency-ready ticket task in an isolated worktree using a small design sketch, behavior-first tests, the appropriate task playbook, and bounded verification. Use for the work stage after ticket planning.

Prepare an approved ticket for human submission by updating durable knowledge, summarizing scope and verification, teaching material flows with progressive visuals, and providing the merge handoff. Use only after review approval. Unless codeHost.pullRequests is false, commit the approved work, push the ticket branch, and create or update the pull request. It never merges.

Prove a task against the real artifact before calling it done. Use during factory-work, before ready_for_review, and whenever a check only shows that code compiles or a file changed.

Layered technical-writing standard: Diátaxis structure, Google developer style sentences, STE instruction rules, Global English syntax. Use for /technical-writing or when writing or reviewing docs, RFCs, readmes, PR descriptions, or commit messages.

Cut AI tells from any writing. Must always apply.