CLAW AGENT
AI agent orchestration harness — a Rust/Tauri desktop app for managing multi-agent teams, mission planning, session lifecycle, health monitoring, and Discord integration. The control plane for autonomous AI workflows.
WHAT IT DOES
KEY FEATURES
Mission planner
Define objectives, break into sub-tasks, assign to agent teams. Track progress through a Kanban-style interface.
Multi-agent teams
Spawn, monitor, and coordinate multiple AI agents with distinct capabilities. Agents communicate through structured message passing.
Session management
Full lifecycle tracking for agent sessions: creation, execution, checkpointing, and teardown. Persistent session state across restarts.
Health monitoring
Real-time CPU, memory, and process monitoring. Built-in doctor system diagnoses issues and suggests fixes.
Discord integration
Live feed of agent activity and mission updates pushed to Discord channels for remote monitoring.
Face recognition
FaceManager component for identity-aware agent interactions.
DuckDB persistence
49-table intelligence layer storing facts, corrections, embeddings, conversation history, and semantic code indexes.
CASE STUDY
ABOUT THE PROJECT
Desktop-native AI orchestration harness built with Rust (Tauri v2) and Svelte 5. Designed to coordinate multiple AI agents working as teams on complex, multi-step missions.
Core Systems: - Mission planner — Define objectives, break into sub-tasks, assign to agent teams. Track progress through a Kanban-style interface. - Multi-agent teams — Spawn, monitor, and coordinate multiple AI agents with distinct capabilities. Agents communicate through structured message passing. - Session management — Full lifecycle tracking for agent sessions: creation, execution, checkpointing, and teardown. Persistent session state across restarts. - Health monitoring — Real-time CPU, memory, and process monitoring. Built-in doctor system diagnoses issues and suggests fixes. - Discord integration — Live feed of agent activity and mission updates pushed to Discord channels for remote monitoring. - Face recognition — FaceManager component for identity-aware agent interactions. - DuckDB persistence — 49-table intelligence layer storing facts, corrections, embeddings, conversation history, and semantic code indexes.
Architecture: Rust backend handles agent process management, IPC, system health, and file operations. Svelte 5 frontend provides the operator interface. DuckDB for local persistence. Ollama for embeddings and semantic search. Designed for power users who need to orchestrate AI at scale.
PROJECT DETAILS
ARCHITECTURE HIGHLIGHT
Offline-first with RxDB — works without internet, syncs via Server-Sent Events when back online.
BY THE NUMBERS
- 117 Svelte components
- 5 role-based dashboards
- Multi-location support
- PWA — iOS & Android
PREVIEW
SCREENSHOTS
POS Transaction Screen
screenshot coming soon
Owner KPI Dashboard
screenshot coming soon
Kitchen Display System
screenshot coming soon
Floor Plan Editor
screenshot coming soon
MORE WORK
OTHER PROJECTS
Tattoo Tide
Full-stack tattoo studio platform — 3D body placement selector, AI-powered admin tools, online booking flow, and a portfolio gallery. This site is built on it.
WTFPOS
Offline-first restaurant POS with QR self-ordering, a live kitchen display, floor plan editor, and multi-location inventory. Built for a real BBQ restaurant chain with Svelte 5 and RxDB.
LET'S WORK TOGETHER
HAVE A PROJECT IN MIND?
Looking to build something real — from prototype to production.
GET IN TOUCH →