Trellis

Why we included this project

A coding agent starts from scratch on every session, with no memory of your project, your conventions, or your team's requirements. Trellis fixes that by persisting specs, tasks, and memory directly into your repository, so coding agents like Claude Code or Codex work to your engineering standards from the first run. It works alongside the tool you already use rather than replacing it, which matters for a developer or small team that has watched an agent re-decide the architecture session after session. If your agents keep drifting from house practices between runs, Trellis gives you a concrete way to codify how they should behave in your codebase.