Plugin

trinity

Trinity is a three-agent harness for Claude Code that turns a one-to-four-sentence requirement into a reviewed pull request. /trinity:run spawns Planner (opus) → Generator (sonnet) → Evaluator (sonnet) inside an isolated git worktree, iterating until the Evaluator returns PASS or the iteration cap is hit, then pushes the branch and opens a PR. The split forces an independent reviewer that only reads files and the diff, eliminating the "graded its own homework" failure mode of single-agent loops. UserPromptSubmit and SubagentStop hooks enforce a clean preflight and structured run logs.

• Published source
• Repository

Analysed

11 October 2026

Last observed

12 October 2026

What the code does

• Runs an AI coding agent unsupervised Code that starts an AI coding agent with its approval prompts switched off.

Having a capability is not wrongdoing; many tools need these to do their job. It tells you what to check before you install.

This is public metadata, not a safety certification or complete assessment. Static analysis cannot establish what a tool does at runtime.