Plan

Breaks down goals into concrete, trackable issues with implementation details and test targets. Built around the idea that you shouldn't ship slices in parallel unless you've proven they won't collide on migrations, contracts, or CLI surface. Enforces a stale-scope gate if you're working from a week-old bead or handoff doc, loads compiled prevention rules from prior failures, and won't let you decompose until you've run a baseline audit to count files and LOC. Outputs markdown plans plus bd issues, scales detail level by complexity (minimal for 1-2 issues, deep for 7+), and integrates with the research and pre-mortem skills. The wave-validity check and symbol-level implementation specs are the real workhorses here.

Install to Claude Code

npx -y skills add boshu2/agentops --skill plan --agent claude-code