Integrated full stack for the TreeHacks build:

• api/ - FastAPI + Elasticsearch backend
• frontend/ - Next.js frontend
• fetch-agents/ - Fetch.ai uAgents track (ASI:One, Agentverse, optional RunPod Flash)

1. Backend env:
   • Copy api/.env.example to api/.env
   • Set ELASTICSEARCH_URL and ELASTICSEARCH_API_KEY
2. Install deps:
   • cd frontend && npm install --legacy-peer-deps
   • cd ../api && python3 -m venv .venv && .venv/bin/pip install -r requirements.txt
3. Start both from repo root:
   • ./scripts/dev.sh

Frontend: http://127.0.0.1:3000
API: http://127.0.0.1:8000

The complete Fetch.ai implementation is in fetch-agents/ (copied from treehacks2026-36), including:

• Orchestrator / Coordinator / Specialist chat flow
• Router / Curator / Expert / Stuck-agent Q&A marketplace
• Payment protocol flow (FET)
• Optional RunPod Flash enrichment via runpod_assist.py

Setup:

1. cd fetch-agents
2. python3 -m venv .venv && source .venv/bin/activate
3. pip install -r requirements.txt
4. cp .env.example .env
5. Set at least AGENTVERSE_API_KEY in .env
6. Run agents as documented in fetch-agents/README.md and fetch-agents/DEMO.md

Important local integration default:

• fetch-agents/.env.example now points HACKOVERFLOW_API_URL to http://127.0.0.1:8000 so Fetch digest/context calls target this local backend.
• Fetch agent default ports are 8100-8106 to avoid conflicts with the backend on 8000.