Hive Mind

🧠 hive-mind

**One memory. Every AI coding tool. Every machine.**

📖 **Full docs: [tuahear.github.io/hive-mind](https://tuahear.github.io/hive-mind)**

You teach Claude Code something useful today — a project quirk, a shell preference, a debugging insight. Tomorrow you reach for Codex, or try Qwen on a side task, or open the same project on another machine. They all start from zero. You re-explain everything, every time.

hive-mind fixes that. A single `~/.hive-mind/` hub per machine holds one canonical memory. You attach each AI tool to the hub — Claude Code today; Codex, Qwen, and Kimi adapters planned — and the hub keeps each tool's native memory files in sync without forcing any of them to change. Claude Code keeps writing to `~/.claude/CLAUDE.md`, Codex keeps reading `~/.codex/AGENTS.md`, and so on. The hub just makes sure they all see the same thing. Same hub on a second machine pulls from the same private git repo you control, so whatever you teach Claude on your laptop shows up in Codex on your desktop.

   ┌─────────────────────────────────────────────────────┐
   │  ~/.hive-mind/ hub (one per machine, one git repo)  │
   │                                                     │
   │   content.md ── skills/ ── projects//           │
   │                                                     │
   │  ├─ attached ──▶ ~/.claude/   (CLAUDE.md, skills)   │
   │  ├─ attached ──▶ ~/.codex/    (AGENTS.md, ...)      │
   │  ├─ attached ──▶ ~/.qwen/     (QWEN.md, ...)        │
   │  └─ attached ──▶ ~/.kimi/     (KIMI.md, ...)        │
   └──────────────────────┬──────────────────────────────┘
                          │ push/pull
                          ▼
            ┌───────────────────────────────┐
            │  your private memory git repo │   ← any git remote you control
            └───────────────────────────────┘
                          ▲
                          │
       same hub on a second machine pulls the same content down

Only using one