
mnemo
Your AI conversations are your decision journals
What is mnemo?
mnemo indexes your AI coding conversations into one searchable database. If you use Claude Code, Cursor, OpenCode, or any of 12 AI coding tools, mnemo reads their native storage formats and makes everything searchable in ~0.8 seconds.
Your AI chats aren't just conversations, they're your decision journal. Every architecture choice, debug session, and trade-off you made at 2am. mnemo makes all of it findable.
Quick Start
brew install Pilan-AI/tap/mnemo
mnemo index
mnemo search "authentication flow"
That's it. First run auto-detects your installed tools and indexes everything. Subsequent runs are incremental.
How it works
12 format adapters parse native storage directly: JSONL, JSON, SQLite. No exports, no copy-paste, no API keys.
Single SQLite database at ~/.mnemo/mnemo.db with FTS5 full-text search. Results ranked by relevance, recency, and match density.
Session-grouped results: when you search "auth flow," you get the 5 most relevant conversations, not 50 scattered messages.
Auto-context for Claude Code: your AI remembers what you discussed last week via MCP and Hooks integration.
Tech Stack
Pure-Go SQLite (modernc.org/sqlite) - single binary, cross-compiles everywhere, no CGO
WAL mode - concurrent reads never block
BM25 + temporal decay - last week = 80% weight, last month = 40%
Zero runtime dependencies - works on first run, everything local
Supported Tools
Claude Code, OpenCode, Cursor, Crush, Gemini CLI, Codex, Amp, Kiro, Cline, Roo Code, Kilo Code, Antigravity.
Why Build This?
I had 89,037 messages across 12 tools. Completely unsearchable. Needed to find "why I chose JWT vs sessions" from 3 weeks ago. Built mnemo to scratch my own itch.
Now my AI coding sessions are my searchable decision journal. Fully local, always fast.
GitHub: github.com/Pilan-AI/mnemo
Built by @0xRaghu | Part of Pilan
எண்ணென்ப ஏனை எழுத்தென்ப இவ்விரண்டும் கண்ணென்ப வாழும் உயிர்க்கு.
"Numbers and letters - these two are the eyes of all who live."
- Thiruvalluvar, Tirukkural 392
About
One day I calculated that I had 89000 messages across 12 coding tools in my history and all of them were getting lost! I noticed that these were rich decision journals and started indexing and making them searchable

Comment