BrightBighead1/supercharged-metagpt
0
๐ง Supercharged MetaGPT
AI-powered multi-agent software development team. One requirement โ full software company output.
Agent Team
Integrated Superpowers
Features
- Real-time streaming โ watch agents think and act live via SSE
- 10+ LLM providers โ OpenAI, Anthropic, Ollama, Gemini, Zhipuai, Fireworks, Azure, HuggingFace
- Project download โ get all generated files as ZIP
- Code review mode โ adds QA step for higher quality
- REST API โ programmatic access for automation
- S3 storage โ optional cloud storage backend
- Mermaid diagrams โ architecture diagrams in output
- Git repo management โ proper version control for generated projects
- Storage auto-cleanup โ prevents disk from filling up
- API key validation โ test keys before running
- Configurable via UI โ change LLM provider, model, settings from the browser
LLM Configuration
Set environment variables or use the Settings tab in the UI:
Free option: Set only HF_TOKEN and the Space uses HuggingFace Inference API (Qwen2.5-72B-Instruct by default).
REST API
# Run full MetaGPT pipeline (streaming)
curl -X POST https://brightbighead1-supercharged-metagpt.hf.space/api/run \
-H "Content-Type: application/json" \
-d '{"requirement": "Build a todo app with user auth, REST API, and tests", "mode": "full"}'
# Quick mode (single engineer)
curl -X POST https://brightbighead1-supercharged-metagpt.hf.space/api/run \
-H "Content-Type: application/json" \
-d '{"requirement": "Write a Python function to sort a list", "mode": "quick"}'
# Check status
curl https://brightbighead1-supercharged-metagpt.hf.space/api/status
# Get team info
curl https://brightbighead1-supercharged-metagpt.hf.space/api/team
# Test API key
curl -X POST https://brightbighead1-supercharged-metagpt.hf.space/api/test-api-key \
-H "Content-Type: application/json" \
-d '{"api_key": "sk-...", "llm_type": "openai"}'Sleep Behavior
The Space stays awake for 48 hours after last use on free hardware. Any visit to the URL wakes it up automatically.
