Tribunal Install

Install and initialise Tribunal — the Claude Code quality enforcement plugin. Use when setting up a new project for AI-assisted development, when a project lacks TDD enforcement, secret scanning, or Claude Code quality gates, or when asked to "add quality enforcement", "set up tribunal", or "protect this project". Works with pip, uv, npm, and brew.

Tribunal — Install & Initialise

Tribunal enforces code quality when Claude Code writes files: TDD enforcement, secret scanning, context monitoring, Agent Teams quality gates, and a live dashboard.

Install

Python / uv (recommended)

bash

pip install tribunal
# or
uv add tribunal

npm

bash

npm install -g tribunal

Homebrew

bash

brew install thebotclub/tap/tribunal

Verify install:

bash

tribunal --version

Initialise a project

Run in your project root (where .claude/ lives or will live):

bash

tribunal init

The wizard asks:

• Primary language (Python / TypeScript / Go / Other)
• TDD enforcement level (Strict = block on fail / Advisory = warn / Off)
• Secret scanning (On/Off — default On)
• Spec workflow (Yes/No — default Yes)
• Context monitor (On/Off — default On)

For non-interactive / agent use:

bash

tribunal init --non-interactive --project-dir /path/to/project

Outputs .claude/tribunal.json and merges into .claude/settings.json.

Verify setup

bash

tribunal doctor

Checks all hooks, Python deps, worker service, and .claude/settings.json links. Exits 1 if anything is broken — safe to use in CI.

Quick install + init (one-liner for agents)

bash

pip install tribunal && tribunal init --non-interactive && tribunal doctor

Install a quality pack

bash

tribunal list-packs          # see available packs
tribunal install python-strict
tribunal install go-tdd
tribunal install nextjs-quality

Workflow

Once installed, Tribunal hooks into Claude Code automatically:

1. Every file Claude writes → file_checker runs (secrets, language quality)
2. Every test run → tdd_enforcer checks pass/fail
3. Context usage → context_monitor warns before window fills
4. Agent Teams → teammate_idle + task_completed gate sub-agent output
5. Dashboard → tribunal/ui/viewer.html shows live hook feed

Run standalone quality check (no Claude Code needed)

bash

tribunal ci                        # check changed files
tribunal ci --json                 # machine-readable output
tribunal ci --coverage-threshold 80

Troubleshoot

bash

tribunal doctor              # health check with fix suggestions
tribunal status              # show current config + audit summary

Full docs: https://tribunal.dev · GitHub: https://github.com/thebotclub/tribunal