# mindfold-ai/Trellis

The best agent harness.

Repository: https://github.com/mindfold-ai/Trellis
Canonical: https://ross.abutalabs.com/products/trellis
Homepage: https://docs.trytrellis.app
Language: TypeScript
License: AGPL-3.0
License Family: copyleft
Topics: agentic-coding, ai-workflow, codex, harness, claudecode
Last push: 2026-08-21T02:13:13+00:00

## Health v2 (maintenance only)
Score: 59/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 98, release rhythm 35, longevity 15
- inputs: {"age_days": 219, "days_push": 13, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 14252, forks 797 (observed 2026-08-28T04:11:06.618773+00:00)

## What it is
Trellis is an open-source engineering framework and agent harness for AI coding assistants that persists project specs, tasks, and cross-session memory into your repository. It auto-injects curated context into every AI session across 20+ coding platforms (Claude Code, Codex, Cursor, Gemini CLI, and more) so agents follow your team's conventions and workflow.

## Use cases
- give AI coding agents persistent memory of my project conventions
- inject project specs automatically into Claude Code or Codex sessions
- enforce a spec-driven workflow on AI coding assistants
- share coding standards and tasks with a team using AI agents
- manage task lifecycle and context for sub-agents
- make any coding agent follow my engineering standards
- stop AI assistants from starting every session from scratch

## When to choose
- you use AI coding assistants (Claude Code, Codex, Cursor, etc.) and want consistent, convention-following output across sessions
- your team needs shared, versioned specs and task memory persisted in the repo
- you want automatic context injection and workflow enforcement rather than manually maintained rules files like .cursorrules or CLAUDE.md
- you work across multiple AI coding platforms and want one portable framework

## When to avoid
- you want a fully autonomous coding agent rather than a harness that scaffolds and constrains existing agents
- your project is small enough that a single rules file suffices
- you need a permissive license - Trellis is AGPL-3.0
- you cannot install Node.js 18+ and Python 3.9+

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, prompt-engineering, developer-tools, cli, workflow-automation
- domain: developer-tools, large-language-models
- platform: python, cli, cross-platform, windows
- tags: agentic-coding, ai-coding-assistants, spec-injection, claude-code, codex, cursor, agent-harness, context-management, npm-package, ai-agents, automation, nodejs, macos, linux

## Member repositories
- mindfold-ai/Trellis (main) score 59

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:06.618773+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:12:41.839927+00:00, confidence not recorded.
  - readme: https://github.com/mindfold-ai/Trellis (fetched 2026-08-28T04:11:06.618773+00:00, sha 2671b2de5a49)
  - homepage: https://docs.trytrellis.app (fetched 2026-08-29T08:06:36.448098+00:00, sha 91795d477f06)
  - site_page: https://docs.trytrellis.app/changelog/v0.6.16 (fetched 2026-08-29T08:06:36.450892+00:00, sha a2a3909a79c4)
  - site_page: https://docs.trytrellis.app/start/install-and-first-task (fetched 2026-08-29T08:06:36.452760+00:00, sha 23848425447e)
  - site_page: https://docs.trytrellis.app/advanced/appendix-f (fetched 2026-08-29T08:06:36.455091+00:00, sha 3347506cc63c)
- Data as of 2026-08-30T08:39:29.467469+00:00.