Scheduled for Thursday, 00:01 PT. Before it goes live, the decision I'm least sure about: "Fix from here" opens a full agent session rather than filing an issue or opening a draft PR directly. The session edits an isolated checkout, runs, verifies, and only then pushes a branch and opens the pull request a person still merges it. Slower than a one-click auto-PR, on purpose: I wanted a point where a human reads what actually changed before it enters a checkout, not after. Genuinely unsure this is the right tradeoff for everyone. If you've built something in this space where would you make the agent commit faster, and where would you make it slower?
Most code tools read one repository at a time. Celmis indexes a set of them into one symbol graph, so a question can cross a boundary: ask which service still calls a function whose signature changed, and the answer comes back with file:line citations from a repository you never cloned.
Same index, read many ways: PR review that sees cross-repo callers, dependency audit, CycloneDX SBOM, and an MCP server with 18 tools.
Self-hosted, docker compose. Your model key, or Ollama locally. AGPL-3.0.
