plan-flow

# Plan Flow ## Overview Use this skill when the user needs: - A complete analysis of duplicated/redundant design in a codebase. - A detailed TODO plan with explicit file-level steps. - Strict execution evidence: change one step, test it, then update plan status. This skill is repository-agnostic. It defines how to analyze and plan, not only what was done in one specific repo. ## Core Workflow (Analyze -> Plan -> Execute) 1. Establish scope and constraints. - Confirm target directories/modules and out-of-scope areas. - Capture compatibility requirement, risk tolerance, and testing expectations. - Record baseline (`git status --short`, current branch, known blockers). 2. Run structured redundancy analysis first. - Build an inventory of architecture anchors: - Domain models and schemas - Factory/registry entry points - HTTP/storage/cache/logging abstractions - Route/service/provider adapters - Identify duplicate/redundant candidates with evidence: - Same concept, multiple conflicting definitions - Same responsibility, parallel implementations - Exported but unconnected modules - Dead/legacy paths still affecting readability - For each finding, record: - Exact files and symbols - Call path or usage evidence - Risk if changed - See `references/analysis-playbook.md`. 3. Prioritize and convert analysis into executable plan. - Score each finding by impact/effort/risk/confidence. - Group into phases (`P0`, `P1`, `P2`) and sequence low-risk/high-signal step...