doc-generatorUse this agent to generate or update READMEs, API docs, inline comments, and changelogs. Use when: User asks to write or update documentation, README, API docs, CHANGELOG, or code comments. Do NOT use when: User needs code exploration (use engineer), test writing (use qa-engineer), or security analy
engineerUse this agent for code exploration, architecture design, refactoring analysis, git operations, and code review. Use when: User asks to explore code, design architecture, refactor, commit/PR, or review changes. Do NOT use when: User needs test generation (use qa-engineer), security audit (use securi
haiku-workerLightweight autonomous task executor for simple/mechanical plan tasks. Uses haiku model for cost efficiency. Use when: Orchestrator needs a cheap worker for tests, docs, formatting, linting, config changes, or file renames. Do NOT use when: Task requires complex reasoning, multi-file architecture, d
qa-engineerUse this agent for test generation, test execution, bug tracking, and deployment validation. Use when: User asks to write tests, run tests, track bugs, or validate deployment readiness. Do NOT use when: User needs code review (use engineer), security scan (use security-scanner), or code exploration
security-scannerUse this agent for security vulnerability scanning including OWASP Top 10, hardcoded secrets, and injection risks. Use when: User asks about security, vulnerabilities, hardcoded secrets, injection risks, or OWASP compliance. Do NOT use when: User needs code review for quality (use engineer), test ge
swarm-workerAutonomous task executor for devloop:run-swarm. Implements a single plan task with fresh context. Do not use for interactive work. Use when: Orchestrator is running swarm mode and needs to parallelize independent plan tasks. Do NOT use when: Interactive work is needed, or for single-task execution (