DaiDocs
github.com/kerneta/daidocs- Category
- Developer Tools
- Rank
- No. 1370Tools index
Previous survey · No. 1333 ·
- Type
- TOOL
- Use case
- Data, Retrieval & Knowledge
- Interfaces
- MCP · Agent Skill / Plugin
- GitHub
- 54 stars
- Latest release
- V5.1.1
- Date
About
DaiDocs is an open plain-text file format (.dai files) for storing AI assistant long-term memory directly on disk, readable by Claude, GPT, Gemini, Cursor, local models, and even grep. It ships with an MCP server and hooks for Claude Code, Claude Desktop, Cursor, Windsurf, and Codex, and reports 83% accuracy on the LongMemEval-S benchmark with GPT-4o and 92% with Claude, using roughly 10x fewer tokens than typical approaches.
Why it made the leaderboard
DaiDocs stores agent long-term memory as plain-text files on disk readable by any model or by grep, avoiding vendor lock-in to a memory service, with claimed 83-92% LongMemEval-S accuracy at a fraction of the token cost.
Tags
Tech Stack
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.