A state-of-the-art, autonomous AI system for end-to-end GitHub repository and project management. Deploy specialized micro-agents to handle code reviews, issue management, CI/CD, security scanning, and documentation—all with minimal human intervention.
- Multi-Agent Architecture: 7 specialized agents working in concert
- Intelligent Code Review: Automated PR reviews with security, quality, and performance checks
- Issue Management: Auto-triage, label, assign, and link related issues
- Branch Operations: Automated merge, rebase, conflict resolution
- Security First: Dependency scanning, secret detection, OWASP compliance
- CI/CD Optimization: Workflow analysis and auto-healingh
- Documentation Generation: Always up-to-date README, API docs, changelogs
- Audit Logging: Full traceability with rollback support
Add autonomous AI code review, issue triage, and security scanning to any repository in seconds:
# .github/workflows/agent.yml
name: Autonomous Agent
on: [push, pull_request, issues]
jobs:
agent:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- uses: labgadget015-dotcom/autonomous-github-agent@main
with:
github_token: ${{ secrets.GITHUB_TOKEN }}
anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
analysis_mode: auto
local_llm_enabled: "true" # route simple tasks to Ollama at $0
severity_threshold: high| Input | Default | Description |
|---|---|---|
github_token |
— | Required. GitHub token for API access |
anthropic_api_key |
— | Anthropic API key (recommended) |
openai_api_key |
— | OpenAI API key (alternative) |
analysis_mode |
auto |
parallel / sequential / auto |
local_llm_enabled |
true |
Route low-complexity tasks to local Ollama at zero cost |
local_llm_endpoint |
http://localhost:11434 |
Local LLM base URL |
complexity_threshold |
medium |
LLM routing cutoff: low / medium / high |
issue_creation_enabled |
true |
Auto-open GitHub issues for critical findings |
severity_threshold |
high |
Minimum severity to trigger issue creation |
Outputs: analysis_summary · issues_created · cost_savings · execution_time
Embed a live GadgetLab score badge in your README — updates every 5 minutes:
[](https://gadgetlab.uk)Replace YOUR_ORG/YOUR_REPO with your GitHub repo path. The score reflects CI status, open issue count, and commit recency.
# Install
pip install -e .
# Configure
cp config/config.example.yaml config/config.yaml
# Edit config.yaml with your GitHub token and preferences
# Run health check on a repository
python -m autonomous_agent health-check --repo owner/repo-name
# Start autonomous monitoring
python -m autonomous_agent monitor --repo owner/repo-name┌─────────────────────────────────────────────┐
│ Orchestrator (Central Brain) │
└──────────────┬──────────────────────────────┘
│
┌───────┴───────┐
│ Event Bus │
└───────┬───────┘
│
┌──────────┼──────────┐
▼ ▼ ▼
┌────────┐ ┌────────┐ ┌──────────┐
│ Health │ │ Code │ │ Security │
│Monitor │ │Reviewer│ │ Scanner │
└────────┘ └────────┘ └──────────┘
▼ ▼ ▼
┌────────┐ ┌────────┐ ┌──────────┐
│ Issue │ │ Branch │ │Workflow │
│Manager │ │Manager │ │Optimizer │
└────────┘ └────────┘ └──────────┘
│
▼
┌──────────┐
│ Docs │
│Generator │
└──────────┘
- Python 3.11+
- GitHub Personal Access Token with repo, workflow, and admin permissions
- Optional: OpenAI/Anthropic API key for enhanced AI features
git clone https://github.com/yourusername/autonomous-github-agent.git
cd autonomous-github-agent
pip install -e .Create config/config.yaml:
github:
token: ${GITHUB_TOKEN}
llm:
provider: openai # openai, anthropic, or local
api_key: ${OPENAI_API_KEY}
model: gpt-4
agents:
enabled:
- health_monitor
- code_reviewer
- issue_manager
- branch_manager
- security_scanner
- workflow_optimizer
- documentation_generator
automation_level: semi-auto # manual, semi-auto, or full-auto
safety:
require_approval_for:
- branch_deletion
- force_push
- workflow_modification
audit_log_retention_days: 90| Agent | Purpose | Key Actions |
|---|---|---|
| Health Monitor | Repository health assessment | Metrics tracking, debt identification, cleanup recommendations |
| Code Reviewer | Automated PR reviews | Quality checks, security scans, inline suggestions, auto-merge |
| Issue Manager | Issue triage & organization | Auto-label, assign, link, close resolved issues |
| Branch Manager | Branch operations | Merge, rebase, conflict resolution, cleanup |
| Security Scanner | Security & compliance | Dependency scanning, secret detection, OWASP checks |
| Workflow Optimizer | CI/CD management | Workflow analysis, failure recovery, optimization |
| Documentation Generator | Documentation updates | README, API docs, changelogs, release notes |
- Human-in-the-Loop: Destructive operations require explicit approval
- Audit Logging: Every action is logged with rollback instructions
- Configurable Automation: Choose your comfort level (manual → full-auto)
- Rate Limiting: Protects against API abuse and cost overruns
# Install dev dependencies
pip install -e ".[dev]"
# Run tests
pytest
# Run with coverage
pytest --cov=autonomous_agent
# Lint
ruff check .
black .
mypy .Contributions welcome! See CONTRIBUTING.md for guidelines.
MIT License - see LICENSE for details.
Built with PyGithub, modern LLMs, and a passion for automation.
Gadget Lab builds AI automation systems for businesses that want to move faster without hiring more people.
This agent is part of a broader stack we run in production — including autonomous n8n workflows, LLM-powered ops systems, and self-healing infrastructure.
If you want this kind of automation working in your business:
- 📧 Get in touch: labgadget015@gmail.com
-
- 💼 What we build: Custom AI agents, workflow automation, GitHub CI/CD intelligence, and ops dashboards
-
-
🎯 Who we work with: UK SMBs and technical teams who want real automation, not just demos
-
"We don't just design systems — we run them. Everything in this repo is live in our own production stack."
Built by Gadget Lab · AI Automation Architect · UK
-
Built with PyGithub, modern LLMs, and a passion for automation.