skillfed

trinity-agent

Three minds, one context — Multi-agent AI orchestrator for Claude Code, Codex, and Antigravity CLI.

trinity-agent v1.0.1163 182.6K downloads/30d#10,096 on PyPI18
Permissive license MIT Active released

What it is and what it does

Trinity is a multi-agent AI orchestrator that coordinates Claude, Codex, and Antigravity CLI into a single collaborative system. Instead of delegating all work to one AI, Trinity runs structured round-based deliberation where each agent contributes its perspective, reaches consensus on a plan, then distributes tasks according to each agent's specialty—Claude as architect, Codex as implementer, Antigravity as reviewer. The system maintains a shared context file, detects blocking questions that need user input, and automates peer review of generated code before final delivery.

It runs as a CLI tool with an optional interactive Textual-based workbench. You initialize a project with `trinity init`, then either ask one-off questions via `trinity ask` or launch the full TUI with `trinity`. The workflow persists state in `.trinity/` so you can pause, ask follow-up questions, and resume. All three provider CLIs are optional—you can start with just one—but having all three unlocks the full multi-agent deliberation and review flow.

Use it for:

  • Design a system architecture with multiple AI perspectives before implementation, catching blind spots a single model might miss.
  • Generate and review code in one workflow: Claude designs, Codex implements, Antigravity validates—all in one session.
  • Iteratively refine a feature by asking follow-up questions mid-deliberation without losing context or restarting the agents.
  • Automate code review by having non-owner agents peer-review each work package before you accept the changes.
  • Manage complex multi-step projects where different agents' strengths (design, coding, edge-case analysis) are needed in sequence.
  • Prototype faster by running deliberation once to get consensus, then executing the agreed plan without manual task assignment.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Trinity orchestrates multiple AI coding agents (Claude, Codex, Antigravity) in structured round-based deliberation, reaching consensus before distributing tasks to each agent's specialty.

Yes, with conditions. Install if you have at least one of Claude Code, Codex, or Antigravity CLI already set up and want to experiment with multi-agent deliberation for code design and review. The low install friction and active maintenance make it worth trying. However, alpha status means workflows and APIs may change; use it for exploration and prototyping rather than critical production pipelines. If you only have one provider CLI, you'll get single-agent interaction rather than true multi-agent consensus, which reduces the value proposition.

Install

trinity-agent on PyPI

pip

pip install trinity-agent

uv

uv add trinity-agent

poetry

poetry add trinity-agent

Installing trinity-agent

Before you install

Low friction install via pipx or pip with six lightweight runtime dependencies. Active maintenance (last commit 2026-08-03, release 17 days old), though early alpha status (Development Status :: 3 - Alpha) means the API and workflow may still shift.

License in practice

MIT license is permissive—you can use, modify, and distribute Trinity freely in commercial or private projects with minimal restrictions, provided you include the license notice.

Quickstart

pipx install trinity-agent
trinity init
trinity ask "design an authentication system"

# Or start the interactive TUI:
trinity

Requires at least one of Claude Code CLI (claude), Codex CLI (codex), or Antigravity CLI (agy) to be installed and authenticated in the same shell environment; Trinity alone cannot invoke agents without them.

Verify before relying

  • Whether provider CLI session resumption across Trinity restarts actually works as described in the workflow documentation.
  • Performance characteristics when running deliberation with all three agents versus a subset.
  • Stability of the Textual TUI on different terminal emulators and operating systems.
  • Whether the central synthesis step reliably detects when a question is blocking versus when a blueprint is ready.

Package facts

License MIT (permissive)
Python support supports the current Python release (<3.14,>=3.10)
Install friction low — pure-Python wheel
Runtime dependencies 6 — click, prompt-toolkit, rich, textual, tomli-w, tomli
Maintenance actively maintained — 17 days since the last release
Last repo commit
First released
Downloads 182,550/month — #10,096 on PyPI (30-day window, as of 2026-08-14)
Known vulnerabilities none known (OSV.dev, checked 2026-08-14)

Evidence: trinity_agent-1.0.1163-py3-none-any.whl

Keywords: ai, antigravity, claude, codex, deliberation, multi-agent, orchestrator

Development Status :: 3 - AlphaIntended Audience :: DevelopersLicense :: OSI Approved :: MIT LicenseProgramming Language :: Python :: 3Programming Language :: Python :: 3.10Programming Language :: Python :: 3.11Programming Language :: Python :: 3.12Topic :: Scientific/Engineering :: Artificial IntelligenceTopic :: Software Development :: Libraries :: Python Modules

Tags

multi-agent ai orchestratorclaude codex antigravity integrationai agent consensus deliberationcollaborative ai coding agentsai agent task distributionmulti-model code generationagent peer review workflow
multi-agent-orchestrationai-code-generationcli-tool

More Python Modules packages

Further reading