I built Syngraphe because I wanted repository context to stay in the repository, instead of living inside the memory system of one specific AI tool. I was already keeping this kind of context manually in Markdown, so I turned that workflow into a small CLI. Syngraphe keeps architecture, conventions, current state, decisions and history under a `.context/` directory. AGENTS.md stays small and mostly acts as the entry point for coding agents. Everything is plain Markdown committed with Git. Syngraphe adds the useful tooling around it: initialization, integrity checks, context size stats, document helpers, monorepo scopes and a GitHub Action. There is no database, cloud memory or required model. If Syngraphe disappeared, the context would still just be normal files in the repo. It's open source, and I'm mainly interested in feedback from people using coding agents on real projects. Docs: https://syngraphe.dev GitHub: https://github.com/suffro/syngraphe

Syngraphe

Structured repo context behind AGENTS.md, for AI and humans.

Structured repo context behind AGENTS.md, for AI and humans.