daystar7777/agent-work-mem
Persistent shared memory and handoffs across any combination of AI coding agents — Claude Code, Codex CLI, Cursor, Aider — using nothing but markdown.
SUMMARY AI summary by gpt-5-mini
agent-work-mem is a small, file-based collaboration protocol (just Markdown) that gives projects a shared, persistent working memory for multiple AI coding agents. It’s aimed at developers who run multiple LLM-based coding assistants (Claude, Codex, Gemini, OpenCode, Cursor, Aider, etc.) and need reliable handoffs, audit trails, and cross-session continuity without a daemon or external service. Key features: - One-prompt install creates AIMemory/ with INDEX.md, PROJECT_OVERVIEW.md, PROTOCOL.md, work.log, handoff_*.md, archive/ and cold/ digests. - Fixed read order (INDEX → PROJECT_OVERVIEW → work.log tail) and append-only HOT log with agent identity and capability tags. - AICP handoff files for cross-agent tasks and tmux optional delivery channel. - Tiered storage (hot, warm archive, cold digests) to avoid context bloat.
DETECTED Detected AI stack
AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.
GitHub Topics
Owner
Dates
| Created on GitHub | 2026-04-26 |
| Last push | 2026-05-09 |
| First seen here | 2026-05-09 |
| Last fetched | 2026-05-09 15:42 |