work-workflow

Use this agent to generate a structured Product Brief from ticket requirements. The brief-writer analyzes ticket details and organizes them into a clear document covering problem statement, goals, requirements (P0/P1/P2), constraints, scope, and success metrics. This agent is invoked automatically by the /work workflow during the 3_brief step. <example> Context: The /work orchestrator needs a brief for a ticket user: "Generate a product brief for PROJ-123" assistant: "I'll use the brief-writer agent to structure the ticket requirements into a product brief" <commentary> The brief-writer reads ticket details and produces a structured brief.md that feeds into the spec stage. </commentary> </example>

Per-ticket cleanup agent invoked during the `cleanup` workflow step (between ci and reports). Deletes the feature branch, kills tmux sessions scoped to the ticket id, and archives the cleanup record. Refuses to act unless the PR is verified MERGED. CRITICAL: This agent must NEVER invoke itself via Task tool — do the work directly.

Designs feature architectures by analyzing existing codebase patterns and conventions, then providing comprehensive implementation blueprints with specific files to create/modify, component designs, data flows, and build sequences

Use this agent to analyze code implementations and provide detailed quality assessments and improvement suggestions. This agent excels at reviewing code patterns, identifying bugs and anti-patterns, suggesting optimizations, assessing test coverage, and ensuring adherence to best practices. Examples include analyzing ESLint rules, React components, Node.js services, test files, configuration files, and providing structured reports with actionable recommendations.

Creates semantic commit messages and pushes. Include "autonomous" or "auto-commit" to skip confirmation.

Setup multiple ticket tasks - creates worktrees and runs custom bootstrap scripts

Generate a structured product brief from ticket requirements or a description

Verify browser/UI state using API-first approach with browser fallback

Run QA testing for a specific app using Playwright MCP

Script-driven quality check (code review, tests, requirements verification)

37 hooks across 5 events