farol-team

farol-team / gnap

Public

GNAP — Git-Native Agent Protocol. RFC Draft for git-based agent orchestration. Zero servers.

agent-coordination agent-orchestration agentic ai ai-agents
15
1
100% credibility
Found Mar 13, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
AI Summary

GNAP is a lightweight protocol for coordinating human and AI agents on tasks using a shared Git repository for defining teams, assigning work, tracking progress, and exchanging messages.

How It Works

1
📰 Discover GNAP

You learn about a straightforward way to help your team members and smart AI helpers work together on projects without needing any extra tools or services.

2
📁 Set up shared workspace

You create a simple dedicated folder in your project space to keep track of everyone and all the work.

3
👥 Build your team

You add the names of people and AI helpers to the team list, noting their roles and readiness to contribute.

4
📋 Create and assign tasks

You write out jobs that need doing and say which team members should handle them.

5
🚀 Watch the magic happen

Your team checks the folder often, grabs their tasks, completes the work, and shares updates automatically.

6
💬 Communicate and review

Team members send notes to each other, check results, and keep everything on track.

🎉 Celebrate success

All tasks are finished, with a clear record of who did what, how it went, and smooth teamwork achieved.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 15 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is gnap?

GNAP is a git-native agent protocol—an RFC draft for orchestrating AI agents like OpenClaw or Claude Code directly through a shared git repo, with zero servers or databases. Developers drop a .gnap/ directory into any repo, define teams in agents.json, assign tasks, and let agents pull work, execute via heartbeats, and push results as runs and messages. It solves multi-agent coordination without vendor lock-in, using git history as the audit log for git-based workflows.

Why is it gaining traction?

Unlike AgentHub or CrewAI, GNAP requires no servers, databases, or runtimes—just git push/pull for eventual consistency, offline capability, and built-in cost tracking via runs. The hook is 30-second setup: commit four JSON types (agents, tasks, runs, messages), and any git-capable agent joins, making it dead simple for git-native orchestration. Humans and AIs mix seamlessly, with git log doubling as your compliance trail.

Who should use this?

AI engineers prototyping agent swarms without infra overhead, startup teams blending human CTOs with bots for task pipelines like billing setup or onboarding, or indie devs building composable dashboards on this protocol. It's for ops-heavy roles ditching Python daemons for pure git transport in agent drafts.

Verdict

Try GNAP if you're experimenting with serverless agent protocols—its spec is crisp, with solid comparisons and production use at Farol Labs—but at 11 stars and 1.0% credibility, it's early-stage vaporware until more adopters validate it. Solid docs make it low-risk to prototype, but wait for v4 stability if production-bound.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.