WeftOS Documentation
Guides, references, and examples for the WeftOS AI framework.
WeftOS Documentation
WeftOS is a full-stack AI operating system built in Rust. It has three layers:
- clawft — The agent runtime. Connects to any LLM, runs on any platform, with built-in tool use and skill learning.
- WeftOS Kernel — Process management, constitutional governance, cryptographic provenance, and self-healing supervision.
- ECC — The cognitive layer. A causal knowledge graph that gives agents persistent memory and reasoning.
Start Here
Quickstart
Install and run your first agent in under 5 minutes.
Architecture Overview
Understand how the three layers work together.
LLM Providers
Configure any of 9 supported LLM providers.
Security Model
Defense-in-depth with WASM sandboxing and governance.
By Layer
Agent Runtime (clawft)
- Getting Started — Installation and first run
- Agent Loop — The 7-stage processing pipeline
- Providers — 9 LLM providers including local inference
- Tools — Built-in and custom agent tools
- Channels — Slack, Teams, Discord, web, CLI
- Security — Sandboxing, SSRF protection, capability system
Kernel (WeftOS)
- Kernel Overview — Boot sequence, services, feature flags
- Process Table — PID management and agent lifecycle
- Governance — Constitutional AI with effect vectors
- ExoChain — Cryptographic provenance and audit trails
- Mesh Networking — Encrypted P2P multi-node coordination
Cognitive Layer (ECC)
- ECC Overview — Causal knowledge graph and semantic search
- DEMOCRITUS — The continuous cognitive loop