claudeers.
// Automation & Workflows

tri-agent-orchestrator

Multi-agent orchestration framework & skill coordinating Google Antigravity, Anthropic Claude Code, and OpenAI Codex.

// Automation & Workflows[ cli ][ api ][ web ][ claude ]#claude#automation◷ MIT$open-sourceupdated 10 days ago

Install with your AI

Paste into Claude Code, Cursor, or any agent — it reads the repo and wires the tool into your project.

Install and set up tri-agent-orchestrator (claude-skill project) into my current project.
Found on https://claudeers.com/tri-agent-orchestrator
Repo: https://github.com/abhigyan6/tri-agent-orchestrator
Homepage/docs: —
Detected install method: claude-skill → # copy this skill into .claude/skills/tri-agent-orchestrator/
Category: automation. Platforms: cli, api, web.
Read the repo's README for exact setup and env vars, then install it and wire it into my project.

Claudeers Health Verdict:
unknown; community-verified: false. Confirm the source before running anything.
// or install directly (claude-skill)
# copy the skill dir into your project:
# .claude/skills/tri-agent-orchestrator/   (or ~/.claude/skills/tri-agent-orchestrator/ for all projects)
// or clone
git clone https://github.com/abhigyan6/tri-agent-orchestrator

// compatibility

Platformscli, api, web
Operating systems—
AI compatibilityclaude
LicenseMIT
Pricingopen-source
LanguagePython

Get your FREE $2.50 API credits to access TickAtlas financial data ↗

🛸 Tri-Agent Orchestrator

Multi-agent orchestration framework and universal agent skill coordinating Google Antigravity (agy), Anthropic Claude Code (claude), and OpenAI Codex (codex).


🌟 Why Tri-Agent Orchestration?

No single AI coding assistant is optimal for every phase of software engineering. Complex software requires diverse cognitive strengths:

  • Google Antigravity (agy) excels at deep codebase reconnaissance, multimodal architecture design, complex dependency analysis, and adversarial edge-case auditing.
  • Anthropic Claude Code (claude) excels at nuanced semantic reasoning, high-taste code craftsmanship, domain modeling, and surgical refactoring.
  • OpenAI Codex (codex) excels at rapid execution, test harness generation, terminal build commands, CI/CD pipeline authoring, and deterministic regression fixes.

Tri-Agent Orchestrator connects all three into a synchronized pipeline using standardized inter-agent contracts (HANDOFF.md), preventing hallucination cascades and delivering production-ready, verified code.


🔄 Core Architecture

flowchart LR
    subgraph Phase1["Phase 1: Architecture & Spec"]
        AGY["🪐 Google Antigravity\n(Chief Architect & Auditor)"]
    end

    subgraph Phase2["Phase 2: Core Craftsmanship"]
        CLAUDE["🧠 Anthropic Claude Code\n(Lead Software Craftsman)"]
    end

    subgraph Phase3["Phase 3: Reliability & Ops"]
        CODEX["⚡ OpenAI Codex\n(Test & Deployment Engineer)"]
    end

    AGY -->|"docs/tri-agent/SPEC.md\n+ HANDOFF.md"| CLAUDE
    CLAUDE -->|"Source Code\n+ HANDOFF.md"| CODEX
    CODEX -->|"Verified Tests\n+ GitHub PR"| Production([Production PR])

📊 Agent Specialization Matrix

Agent CLIPrimary RoleCognitive SuperpowerResponsibilities in Workflow
agy (Antigravity)Chief Architect & AuditorMultimodal analysis, dependency graphs, security modeling• System architecture & RFC drafting
• Adversarial edge-case discovery
• Visual UI/UX verification
claude (Claude Code)Lead Software CraftsmanArchitectural taste, clean abstractions, contextual refactoring• Domain logic & API implementation
• Technical debt refactoring
• Clean code patterns & documentation
codex (Codex CLI)Reliability & Ops EngineerFast execution, test generation, build systems, bash tooling• Unit & integration test suites
• Linting & typecheck fixes
• Git commits & PR preparation

🚀 Quickstart in 60 Seconds

1. Clone & Setup

git clone https://github.com/abhigyan6/tri-agent-orchestrator.git
cd tri-agent-orchestrator

2. Verify Your Environment

Check if all three agent CLIs and developer tools are ready:

./bin/tri-agent doctor

Output:

=======================================================
 🛸 Tri-Agent Orchestrator: Environment Doctor
=======================================================

Coding Agents Status:
  ✅ Google Antigravity (agy)          : 1.5.4
  ✅ Anthropic Claude Code (claude)    : 2.1.322
  ✅ OpenAI Codex (codex)              : 0.167.0

Status Summary:
  🎉 All 3 agent CLIs (Antigravity, Claude Code, Codex) are active and ready!

3. Register the Skill Globally

Install or symlink the skill so Antigravity, Claude Code, and Codex can use it natively:

./scripts/install-skill.sh

📋 The Inter-Agent Handoff Contract (HANDOFF.md)

Whenever an agent concludes its phase, it produces or updates HANDOFF.md in the repository root:

# Tri-Agent Handoff Contract

## Metadata
- **Workflow ID:** `wf-20261110-a9f3b1`
- **Active Feature:** `auth-middleware`
- **Current Phase:** `Phase 1: Architecture & Specification`
- **Completed By:** `agy`
- **Next Agent:** `claude`
- **Timestamp:** `2026-11-10T16:00:00Z`

## What Was Done
- Designed RS256 token rotation architecture.
- Outlined CSRF defense and SameSite=Strict cookie invariants.

## Key Files & Artifacts
- Specification: `docs/tri-agent/SPEC-auth.md`

## Critical Constraints & Decisions
- Token expiration: 15 minutes access, 7 days refresh.
- No third-party heavy dependencies; use native Web Crypto or jose.

## Verification Checklist for Next Agent
- [ ] Implement `src/middleware/auth.ts` matching `SPEC-auth.md`.
- [ ] Export `AuthRequest` and `TokenPayload` types.
- [ ] Update `HANDOFF.md` and set Next Agent to `codex`.

🛠️ Tri-Agent CLI Usage

The repository includes a dedicated CLI helper (bin/tri-agent):

# Check system health across all 3 agent CLIs
tri-agent doctor

# Install the skill to ~/.agents/skills, ~/.claude/skills, and ~/.codex/skills
tri-agent install-skill

# Check current handoff status in the current project
tri-agent handoff status

# Initialize a new handoff contract
tri-agent handoff create --feature "rate-limiter" --completed agy --next claude

# Run an automated relay workflow
tri-agent run --workflow relay --prompt "Build modern auth middleware with JWT rotation"

# Dry run mode (prints planned commands without invoking)
tri-agent run --workflow relay --prompt "Build file upload component" --dry-run

📂 Repository Structure

tri-agent-orchestrator/
├── SKILL.md                            # Universal Agent Skill definition
├── README.md                           # Documentation & architecture guide
├── LICENSE                             # MIT License
├── init-repo.sh                        # Git initialization and GitHub push script
├── bin/
│   └── tri-agent                       # Executable CLI orchestrator
├── lib/
│   ├── agents.py                       # CLI invocation wrappers for agy, claude, codex
│   ├── orchestrator.py                 # Core workflow coordination & handoff logic
│   └── templates.py                    # Contract & specification templates
├── workflows/
│   ├── 01-relay-pipeline.md            # Research -> Build -> Verify pipeline
│   ├── 02-parallel-debate.md           # Multi-agent architecture consensus
│   └── 03-redteam-audit.md             # Implementation & adversarial red-team audit
├── templates/
│   ├── HANDOFF.template.md             # Handoff schema template
│   └── AGENT_SPEC.template.md          # Architectural spec schema template
├── examples/
│   ├── auth-middleware-walkthrough.md  # Auth middleware walkthrough
│   └── file-upload-spec-walkthrough.md # File upload walkthrough
└── scripts/
    └── install-skill.sh                # Cross-agent skill installation script

🤝 Prompting the Agents

Once the skill is active, you can invoke your workflows naturally:

In Antigravity:

"Run the tri-agent workflow to research and build a modern auth middleware for my app."

In Claude Code:

"Read HANDOFF.md and implement the components outlined in docs/tri-agent/SPEC.md."

In OpenAI Codex:

"Check HANDOFF.md, write unit and edge-case tests for the new auth middleware, and verify they all pass."

📄 License

Distributed under the MIT License.

// faq

What is tri-agent-orchestrator?

Multi-agent orchestration framework & skill coordinating Google Antigravity, Anthropic Claude Code, and OpenAI Codex.. It is open-source on GitHub.

Is tri-agent-orchestrator free to use?

tri-agent-orchestrator is open-source under the MIT license, so it is free to use.

What category does tri-agent-orchestrator belong to?

tri-agent-orchestrator is listed under automation in the Claudeers registry of Claude-compatible tools.

6 views
★ 22 stars
unclaimed
updated 10 days ago

// embed badge

tri-agent-orchestrator on Claudeers
[![Claudeers](https://claudeers.com/api/badge/tri-agent-orchestrator.svg)](https://claudeers.com/tri-agent-orchestrator)

// retro hit counter

tri-agent-orchestrator hit counter
[![Hits](https://claudeers.com/api/counter/tri-agent-orchestrator.svg)](https://claudeers.com/tri-agent-orchestrator)

// reviews

// guestbook

0/500

// related in Automation & Workflows

🔓

The agent that grows with you

// automationNousResearch/⟨Python⟩★ 264,763◷ MIT[ claude ]
🔓

The API to search, scrape, and interact with the web at scale. 🔥

// automationfirecrawl/⟨TypeScript⟩★ 199,533◷ AGPL-3.0[ claude ]
🔓

🌐 Make websites accessible for AI agents. Automate tasks online with ease.

// automationbrowser-use/⟨Python⟩★ 121,235◷ MIT[ claude ]
🔓

Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop

// automationLeonxlnx/⟨JavaScript⟩★ 110,676◷ MIT[ claude ]
→ see how tri-agent-orchestrator connects across the ecosystem