SAIPEN Uses Markdown Files to Give AI Coding Agents Persistent Memory
vactower · reddit · 2026-07-22
The author describes the classic problem with AI coding assistants: once a chat is closed, the agent forgets the project and work has to be re-explained from scratch.
To fix that, they built SAIPEN, a plain-text protocol based on markdown files stored in the project. It records:
- the current phase of work
- the to-do board
- a running decision log
Any agent that can read files can pick up the state and continue without model-side memory. The author says they are still actively testing it and links the project on GitHub.
More from coding & agent
- After CrewAI and Hermes-Agent, a creator is building AI “intelligence layers” for clients — alexcovo_eth · 2026-07-22
- Building 3D Game Core Loops with Claude Fable and Three.js Agent Skills — nptacek · 2026-07-22
- Fable users demo a procedural dog simulation built with three.js agent skills — willdepue · 2026-07-22
- FDE: The $1M/Year AI Job and a 30-Day Roadmap to Master It — blaizedsouza · 2026-07-22
- Fireworks says Kimi K3 handles 72–96% of agent traffic at up to 50x lower cost — eliebakouch · 2026-07-22
- AutoIndex lifts CRUMB recall by 8.4% without changing retrievers or embeddings — mrdrozdov · 2026-07-22