pm

Complete a task - supports worktree and in-session workflows

Audit task definitions for quality and completeness (not code review - see /pm:review for that)

Interactive design session to create a new task

Digest an existing design document into individual tasks

Create new tasks from code review suggestions

Comprehensive automated code review that populates REVIEW.md. Use when the user says "full code review", "audit implementation", "review all changes", "populate REVIEW.md", or needs a thorough review of task implementation against acceptance criteria. Runs in isolated context.

Review code for quality, bugs, and adherence to project conventions. Use when the user says "review this code", "check my changes", "code review", or needs a focused review of specific task changes. Lighter than code-auditor - for individual task reviews rather than comprehensive audits.

Analyze task dependency graphs across repos and submodules. Use when the user says "analyze dependencies", "check dependency graph", "find circular dependencies", "show task ordering", "detect orphaned tasks", or needs comprehensive dependency analysis for planning. Returns a detailed report with graph visualization, cycle detection, orphan detection, status computation, and optimal task ordering.

Comprehensive audit of ALL tasks in tasks/ directory. Use when the user says "audit all tasks", "full task review", "backlog maintenance", "find orphaned tasks", or needs a thorough scan of the entire task backlog. Runs in isolated context for large reviews.

Review code for quality, bugs, and best practices. Use when the user says "review this code", "check my implementation", "code quality", "is this code ready", or asks for feedback on specific code. For comprehensive automated reviews that populate REVIEW.md, use the code-auditor agent instead.

Review a specific task or few tasks for quality before implementation. Use when the user says "review this task", "check task quality", "is this task ready", "validate task MKT-003", or is working on a task and wants inline quality feedback. For comprehensive backlog audits, use the task-auditor agent instead.

4 hooks across 3 events