agent.md

Review any artifact and output results in a Retrospective board format — what is well, what is bad, what should be improved, what to change. Use when the user says "review", "critique", "evaluate", "what's wrong with", "compare", or passes a skill, rule, doc, spec, code, diff, or pull request for assessment. Works on a single artifact, an artifact against a reference, a diff or PR, or two artifacts side by side.

Compare options and recommend a direction. Use for decision requests like "choose", "which option", "tradeoffs", "recommend", "should we", and option selection with criteria, risks, and reversibility.

Organize material into meaningful groups. Use for classification requests like "categorize", "group", "cluster", "sort", "taxonomy", "organize these", and grouping by criteria, priority, dependency, similarity, or abstraction level.

Explain any knowledge topic simply and accurately. Use for explanation requests like "explain X", "why/how/what is X?", concepts, science, definitions, code, design, architecture, and walkthroughs.

Investigate local repository, document, and attached-artifact context. Use for local investigation requests like "investigate", "find where", "understand this repo", "trace", and local-context research; do not use for web search.

Manage active work across people, agents, tasks, dependencies, blockers, and handoffs. Use for coordination requests like "manage this work", "lead this", "assign", "delegate", "track blockers", "status", "handoff", and multi-workstream execution.

Sequence work before execution. Use for planning requests like "break this down", "roadmap", "approach", "milestones", "how should we proceed", migration planning, rollout planning, and scoped next steps.

Preserve durable project facts, decisions, and useful observations in memory files. Use for memory requests like "remember this", "save context", "record a decision", "update memory", or preserving a project fact.

Generate high-leverage questions and clarify missing context. Use for question-generation requests like "what should I ask", "right questions", "what are we missing", "clarify this", and ambiguous requests blocked by unknowns.

Diagnose when skills, rules, workflows, docs, evals, or memory conventions are mismatched to current conditions — triggered by failures, friction, user feedback, outdated assumptions, or changed constraints — then name the smallest change needed and route to the skill or workflow that should update it. Use when the user says "adapt based on this", "what should change after this?", "this keeps happening", "this failed, what should change?", "the workflow no longer fits", "the constraints changed", or asks what skill, rule, doc, eval, memory, or process should change.

Use before installing, updating, or trusting a skill from any source. Audits SKILL.md, permissions, dependencies, prompt-injection patterns, network behavior, exfiltration risk, and bundled resources, then returns a severity-based install verdict.

Work through ambiguous problems before a firm output shape is warranted. Use for reasoning requests like "reason through", "think through", "brainstorm", "help me frame this", "let's work through this", and messy problem statements.

Build or modify backend implementation. Use for API routes, services, middleware, workers, persistence, validation, authorization, configuration, observability, and backend behavior tests.

Generate or modify database code: schemas, DDL, SQL queries, migrations, analytics SQL, indexes, stored procedures, and dialect-specific database scripts.

Generate production-ready frontend code. Use for components, pages, routes, client state, forms, styling, accessibility, performance, PWA behavior, and data visualization.

Generate or revise automated tests. Use for E2E/browser, API/contract, integration, and load/performance test requests.

Use when writing or improving CLI-agent rules, custom instruction files, AGENTS.md, CLAUDE.md, Cursor rules, Copilot instructions, or modular `rules/<name>.md` files.

Create, edit, review, evaluate, package, and optimize skills. Use when users ask to create a skill, revise skill instructions, review a skill, run skill evals, benchmark skill performance, package a skill, or optimize a skill description for trigger accuracy.

Contract-first API design. Use when asked to design an API contract, OpenAPI/Swagger spec, AsyncAPI event contract, GraphQL schema, endpoints, schemas, request/response shapes, or "what should the API look like?" Produces an interface spec before implementation.

Handle Git branch naming, branch actions, commit-message drafting, and committing staged changes. Use for Git workflow tasks, branch review, commit requests, and diffs needing commit messages.

Review code changes, diffs, pull requests, branches, or patches. Use for review findings covering correctness, regressions, security, performance, and test gaps.

Use whenever the user asks to write or revise a PRD, product requirements, product brief, feature requirements, product scope, launch requirements.

Write product specifications and requirements documents. Use for tech specs, design docs, TDDs, functional or non-functional requirements, data contracts, UI specs, release specs, handoff docs, and system behavior.

Write technical docs. Use for READMEs, API docs, endpoint references, routine or on-call runbooks, operational procedures, changelogs, and release notes. Use for operational, reference, and release communication docs rather than product or specification docs.

Write user stories with acceptance criteria and developer tasks. Use for story writing, Jira/Linear/GitHub tickets, task breakdowns, story points, and story-level sprint planning.