How yyjeqhc/webcodex works
WebCodex lets AI clients operate on repositories and development tools on user-owned machines. The main workflow is client request → Server routing and authorization → Runner execution against a local project → observable job results for review. Desktop and CLI support setup and administration; additional capabilities include project management, browser/computer control, workflow sessions, and audit evidence. The README documents the overall flow, but the sampled source is partial, so internal Server routing links are limited to relationships explicitly supported by the available evidence.
Main entry points: src WebCodex Server · main.rs Runner service · job_manager.rs Job lifecycle · lib.rs Runner registry
Components 17
User interfaces
- Desktop app
apps/desktop/src/App.tsx - CLI
crates/webcodex-cli/src/webcodex_cli/mod.rs - Desktop workflows
apps/desktop/src/features/projects/ProjectsPanel.tsx
Server and control
- WebCodex Server
src - MCP gateway
crates/webcodex-core/src/mcp_gateway.rs - Runner registry
crates/webcodex-runner-registry/src/lib.rs - Persistent records
crates/webcodex-store/src/lib.rs
Runner execution
- Runner service
crates/webcodex-runner/src/main.rs - Job lifecycle
crates/webcodex-runner/src/webcodex_runner/job_manager.rs - Project access
crates/webcodex-runner/src/webcodex_runner/projects.rs - Local tools
crates/webcodex-runner/src/webcodex_runner/dispatch.rs - Browser control
crates/webcodex-browser/src/lib.rs - Computer control
crates/webcodex-computer/src/lib.rs
Setup and evidence
- Environment setup
crates/webcodex-environment/src/engine.rs - Workflow sessions
crates/webcodex-core/src/workflow_session_contract.rs - Audit evidence
src/audit_http/routes.rs
Other components
- AI client
Connections 18
- AI client → MCP gateway: sends MCP requests
- MCP gateway → WebCodex Server: routes requests
- WebCodex Server → Runner registry: dispatches work
- WebCodex Server → Persistent records: reads and writes
- Runner registry → Runner service: coordinates runners
- Runner service → Job lifecycle: manages jobs
- Job lifecycle → Project access: accesses projects
- Job lifecycle → Local tools: executes operations
- Runner service → Browser control: uses browser control
- Runner service → Computer control: uses computer control
- Runner service → Runner registry: reports job updates
- Desktop app → Desktop workflows: presents workflows
- Desktop app → WebCodex Server: configures connection
- CLI → WebCodex Server: administers services
- CLI → Environment setup: configures runtime
- WebCodex Server → Workflow sessions: records workflow
- WebCodex Server → Audit evidence: serves audit evidence
- AI client → WebCodex Server: connects over HTTPS