Trinity Lite Documentation

Trinity Lite is a local AgentOps control plane for cross-vendor CLI AI agents. It connects the tools developers already use and makes their work routable, recoverable, reviewable, and acceptable only with durable evidence.

It gives existing tools such as Codex, Claude Code, Hermes, or your own CLI a shared workflow:

route -> work -> review -> verify -> accept

The default demo uses mock agents, so you can try the full flow before wiring in real commands.

Start Here

python3 -m pip install trinity-lite
trinity-lite doctor
trinity-lite orchestrate "implement a hello-world function"

For source development:

git clone https://github.com/Yomiracle/trinity-lite.git
cd trinity-lite
python3 -m pip install -e ".[yaml]"
python3 -m unittest discover -s tests -v

Core Docs

• Why Trinity Lite?
• Tutorial
• Real Agent Setup
• Agent Capabilities
• Worktree Parallelism Preview
• Architecture
• Operations
• Security
• MCP Server Design
• Product Positioning

Recipes

• Codex
• Claude Code
• Hermes and private Trinity boundaries
• Generic CLI agent

Optional Extras

python3 -m pip install "trinity-lite[yaml]"          # YAML pipeline files
python3 -m pip install "trinity-lite[mcp]"           # MCP server
python3 -m pip install "trinity-lite[agent-skill]"   # agent-skill-system integration

What Trinity Lite Is Not

Trinity Lite is not a hosted agent platform, a model gateway, or a replacement for a full agent framework. It is the local operations and acceptance layer for existing CLI agents.