Trellis is a framework and toolkit designed to standardize how AI coding assistants work within projects. It automatically injects project specifications, conventions, and context into each AI session across 20+ supported platforms including Claude Code, Cursor, Gemini CLI, and GitHub Copilot. This ensures AI-generated code follows team standards rather than improvising from scratch each session.
The tool organizes work around tasks, specs, and project memory stored in the repository. Teams define conventions once in `.trellis/spec/` and Trellis distributes them to every AI session. It supports a four-phase workflow: planning with brainstorming, implementation with auto-checking against specs and tests, and task archival with journal updates. Built with Node.js and Python, it works as a command-line tool that can be initialized in any project.
Momentum
Daily activity over the last 30 days: upvotes, unique visits, new GitHub stars and Hacker News points.