TL;DR: Git0 is a self-hosted Git service that stores everything on S3 and provides APIs designed for autonomous AI agents. If you're building self-improving AI systems, this is the Git infrastructure they need.
# Clone and run
git clone https://github.com/satyamtiwary/git0
cd git0 && docker-compose up -d
# Your Git service is now running!
# Web API: http://localhost:8000
# S3 Console: http://localhost:9001 (MinIO)GitHub is built for humans. But AI agents are different:
- They generate A LOT of code - Traditional Git storage doesn't scale. S3 does.
- They work 24/7 - No disk space alerts at 3 AM. Infinite S3 storage.
- They need APIs, not UIs - Every operation must be programmable.
- They react to events - Real-time streams trigger autonomous improvements.
- They self-improve - Version control for machines learning from their mistakes.
Git0 is built from the ground up for this new paradigm.
from git0_sdk import Git0Client
import litellm
client = Git0Client(base_url='http://localhost:8000/api/v1')
# Agent monitors its own code quality
async def self_improve():
for event in client.events.stream():
if event.type == 'pull_request.merged':
# Analyze what was merged
pr = event.data
diff = client.pulls.get_diff(pr.org, pr.repo, pr.number)
# Learn from the changes
insights = litellm.completion(
model='gpt-4',
messages=[{
'role': 'system',
'content': 'What patterns can we learn from this PR?'
}, {
'role': 'user',
'content': diff
}]
)
# Apply learnings to improve other code
improvements = generate_improvements(insights)
client.repos.create_pr(
title="Self-improvement based on PR #" + pr.number,
changes=improvements
)# Agent that automatically fixes failing tests
async def auto_fix_bugs():
for event in client.events.stream():
if event.type == 'workflow.failed':
# Get the failure logs
logs = event.data.logs
failing_test = extract_failing_test(logs)
# Generate fix
fix = litellm.completion(
model='claude-3',
messages=[{
'role': 'user',
'content': f'Fix this failing test:\n{failing_test}\nError: {logs}'
}]
)
# Push the fix
client.repos.create_commit(
message=f"Auto-fix: {failing_test}",
changes=fix.code
)┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ AI Agents │────▶│ Git0 API │────▶│ S3 Storage │
│ │ │ │ │ │
│ • Code Gen │ │ • FastAPI │ │ • Infinite │
│ • Self-Improve │ │ • Async │ │ • Scalable │
│ • Auto-Fix │ │ • Event Streams │ │ • Durable │
└─────────────────┘ └─────────────────┘ └─────────────────┘
│
▼
┌─────────────────┐
│ Temporal │
│ │
│ • Workflows │
│ • Reliability │
└─────────────────┘
- 🔌 API-First: Every operation available via REST API
- 📡 Event Streams: Real-time SSE for reactive agents
- 🐍 Python SDK: Native library for agent development
- ♾️ Infinite Storage: S3-native means no storage limits
- 🔄 Temporal Workflows: Reliable background processing
- 🚀 5-Minute Setup: Docker Compose and you're running
- 🔐 Self-Hosted: Your code, your infrastructure
- 📦 S3-Compatible: Works with AWS S3, MinIO, Backblaze, etc.
- 🏢 Multi-Org: Isolate agent experiments by organization
- 📝 Full Git: Push, pull, clone - it's real Git
| Operation | GitHub | Git0 | Improvement |
|---|---|---|---|
| Clone 10GB repo | 12 min | 3 min | 4x faster |
| Storage cost/TB/month | $23* | $5 | 4.6x cheaper |
| Max repo size | 100GB | Unlimited | ∞ |
| Concurrent operations | Limited | Unlimited** | ∞ |
*GitHub LFS pricing / **Limited only by S3 throughput
- Docker & Docker Compose
- 4GB RAM minimum
- S3-compatible storage (or use built-in MinIO)
- Configure S3 (or use MinIO for self-hosted)
S3_ENDPOINT=https://s3.amazonaws.com
S3_ACCESS_KEY=your-key
S3_SECRET_KEY=your-secret
S3_BUCKET=git0-repos- Set Security Keys
API_KEY=generate-a-secure-key
JWT_SECRET_KEY=another-secure-key- Launch Services
docker-compose -f docker-compose.prod.yml up -d- Initialize First Agent
from git0_sdk import Git0Client
client = Git0Client(
base_url='https://git0.yourcompany.com',
api_key='your-api-key'
)
# Create org for your AI agents
client.orgs.create('ai-agents', 'Autonomous AI Development')
# Create first repo
repo = client.repos.create(
'ai-agents',
'self-improving-assistant',
description='AI that improves itself'
)We're building the future of AI agent infrastructure. Join us!
- 💬 Discord: Join our community
- 🐛 Issues: Report bugs
- 💡 Discussions: Share ideas
Built with:
- FastAPI - Async Python web framework
- Temporal - Workflow orchestration
- MinIO - S3-compatible object storage
Git0 - Version control for the age of autonomous AI
Because machines need different tools than humans