GEIANT Hive Worker

Turn your laptop into an AI compute node. Earn GNS tokens.

npx @gns-foundation/hive-worker join

That's it. One command. Your device joins the GEIANT Hive swarm, contributes idle CPU/GPU cycles to AI inference jobs, and earns GNS tokens automatically.

What it does

• Generates a persistent Ed25519 identity (stored at ~/.hive/identity.json)
• Detects your hardware: CPU, GPU, RAM, estimated TFLOPS
• Resolves your location to an H3 geospatial cell (Res-6, ~36 km²)
• Registers you in the GEIANT Hive swarm registry
• Starts a llama.cpp RPC server if available (handles AI model shards)
• Sends a heartbeat every 30 seconds to stay active in the swarm
• Disconnects cleanly on Ctrl+C

Commands

# Join the swarm (with optional GNS handle)
npx @gns-foundation/hive-worker join --handle @yourname

# Join without starting the RPC server (observer mode)
npx @gns-foundation/hive-worker join --no-rpc

# Check swarm status and your token balance
npx @gns-foundation/hive-worker status

# Show your identity
npx @gns-foundation/hive-worker whoami

# Disconnect gracefully
npx @gns-foundation/hive-worker leave

Compute contribution

For full compute participation, install llama.cpp:

# macOS (Apple Silicon)
brew install llama.cpp

# From source
git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp && make rpc-server

The worker detects the binary automatically and starts it on port 50052. Without it, you join as an observer node (still registered in the swarm, no inference workloads routed to you).

Revenue split

Every inference job processed by the Hive distributes GNS tokens automatically:

| Slice | Recipient | |-------|-----------| | 60% | Community node operators (you) | | 25% | GEIANT / ULISSY s.r.l. (orchestration) | | 10% | Hydration & Resilience Fund | | 5% | Sovereign Quorum coordinators |

Your share is weighted by trust tier (breadcrumb history) and thermal stability score.

Privacy

• Your Ed25519 private key never leaves your machine (~/.hive/identity.json, mode 600)
• Location is derived from IP geolocation (city-level, no GPS)
• Inference workloads are encrypted in transit (NaCl)
• You can leave and rejoin at any time — your identity and earnings persist

System requirements

• Node.js ≥ 18
• Any OS (macOS, Linux, Windows)
• Internet connection
• Optional: llama.cpp rpc-server for compute participation

Built on GNS Protocol · GEIANT · Whitepaper
Contact: [email protected] · USPTO #63/948,788