0ctx

It records AI‑driven coding interactions directly in a repository, attaching sessions, checkpoints and decisions to branches, commits and workstreams. When a developer switches between supported agents or closes a tool, the stored context is injected automatically, allowing the next session to continue from the same point without manual reconstruction. The system runs locally as a daemon, persisting data in a SQLite graph that is bound to the repo and accessed through a simple CLI, requiring no configuration beyond enabling it for the project.

The design targets developers who use multiple AI coding assistants and need a durable, repo‑centric memory of their conversations, code suggestions and rationale. By capturing each exchange, linking it to the relevant code state, and exposing insights such as checkpoints and summaries, the tool eliminates the loss of context that occurs when tabs are closed or tools are changed.

Its distinctive aspects are the local‑first architecture, zero‑config enablement per repository, and the use of a lightweight SQLite graph combined with the MCP protocol to keep the memory tightly coupled to the project's version‑control history.