neat-freak

To keep agent memory and project documentation synced, you need to reconcile files like CLAUDE.md and README against the current codebase. This process removes outdated content and ensures future agents work from accurate, non-conflicting sources.

Preventing knowledge rot during handoff requires cross-layer alignment of agent memory and project rules. By performing a complete inventory of markdown files and applying deterministic edits, you ensure teammates and agents receive consistent, accurate project knowledge.

Cross-platform agent handoff works across Claude Code, OpenAI Codex, OpenCode, and OpenClaw by applying appropriate memory-path handling. This ensures knowledge consistency and proper documentation reconciliation regardless of the specific AI coding environment.

To clean up outdated and bloated documentation, you must enforce file size constraints and anti-bloat rules. This removes or migrates mis-scoped historical narratives from root rule files, keeping project documentation concise and strictly relevant.

You need to reconcile agent memory and docs during dev milestones, stale-doc reports, conflicting-memory incidents, and teammate or agent handoffs. This end-of-session cleanup guarantees knowledge consistency before context is lost or transferred.