EdgeCoder Documentation
This docs site covers the full EdgeCoder system: a decentralized mesh network for running AI models at the edge.
What EdgeCoder Is
EdgeCoder turns every device into a unified agent that contributes compute to a peer-to-peer mesh. Each node runs a coordinator, worker, and inference service in a single process. Nodes discover each other via gossip mesh and distribute tasks based on capability and availability.
Users interact through a chat-first desktop app (Tauri + Svelte with Monaco editor), external IDEs via an OpenAI-compatible endpoint, or the iOS app with on-device llama.cpp inference and BLE local mesh.
Why EdgeCoder
- No cloud dependency -- all inference runs locally on commodity hardware using Ollama or llama.cpp.
- Permissionless mesh -- any device can join and contribute compute to earn credits.
- Cryptographically auditable -- every transaction is recorded in an append-only ordering chain anchored to Bitcoin via OP_RETURN.
- Privacy-preserving -- task data stays on the executing node; envelope encryption is staged for activation.
Start Here
- System Overview and Benefits
- How EdgeCoder Works
- Architecture Deep Dive
- IDE Integration
- Request Lifecycle Sequences
- Model Provider Abstraction
- Executor Sandbox and Isolation
- BLE Local Mesh
- Model Management
- Deployment Topology
- Public Mesh Operations
- iOS Background Execution & Compute Modes
- Coordinator Discovery and Failover
- Coordinator Federation
- Trust and Security
- Threat Model
- Credits, Pricing, and Issuance
- Settlement Lifecycle
- Bitcoin Anchoring
- API Surfaces
- API Endpoints Detailed
- Runtime Modes
- Environment Variables
How This Site Is Organized
- Guide -- System architecture, unified agent model, desktop app, IDE integration, BLE mesh, and model management.
- Operations -- Deployment topology, mesh operations, iOS scheduling, coordinator federation, and failover.
- Security -- Trust boundaries, threat model, cryptographic stack, envelope encryption, and audit posture.
- Economy -- Credits, dynamic pricing, Bitcoin anchoring, Lightning settlement, and issuance parameters.
- Reference -- API endpoints, environment variables, runtime modes, and source document index.