Running the Mesh

Launch the full DOF-MESH multi-LLM network — from quick CLI start to full Docker deployment.

Option 1 — CLI (quick start)

pip install dof-sdk==0.8.0

python3 core/api_node_runner.py \
  --nodes deepseek-coder cerebras-llama gemini-flash local-agi-m4max \
  --daemon

Do not include local-qwen — it no longer exists. Use local-agi-m4max for local sovereign compute.

---

Option 2 — Docker (full mesh)

git clone https://github.com/Cyberpaisa/DOF-MESH.git && cd DOF-MESH
cp .env.example .env   # add your API keys
docker-compose up -d --build

curl http://localhost:8000/
# → {"status":"ok","agents":8,"crews":11}

Services:

Service	Port	Description
A2A Server	8000	JSON-RPC + REST agent interface
Streamlit Dashboard	8501	Real-time governance monitoring
Telegram Bot	—	/status, /prove, /verify

---

Running a Crew

python main.py
# → Interactive CLI (15 options)

python main.py --mode research --task "Analyze DOF competitors"
# → Research crew: 5-agent pipeline

---

Governance Verification at Startup

dof prove
# → 4/4 PROVEN in ~110ms

dof verify-states
# → INV-1 a INV-4: PROVEN

dof health
# → DOF v0.8.0 — 16/16 components available

---

Environment Variables (minimum)

GROQ_API_KEY=gsk_...         # Required — free
MINIMAX_API_KEY=...           # Primary provider — 1,000 req/day free
EVM_PRIVATE_KEY=0x...         # For on-chain attestations

---

Logs and Audit Trails

All outputs go to JSONL for audit:

logs/traces/         ← RunTrace per execution
logs/experiments/    ← runs.jsonl with aggregated metrics
logs/metrics/        ← governance + supervisor steps
logs/daemon/         ← cycles.jsonl (autonomous daemon)
logs/mesh/           ← nodes.json, messages.jsonl
logs/z3_proofs.json  ← formal verification results

---

Mesh Nodes

Node table, TTL backoff, consensus

Providers & Agents

Provider chains for each role