Self-host microclaw
sdk-v0.11.1Embeddable Rust agent platform with Server, Work, and SDK
- Stars
- 749
- Port
- 10961 (Web UI)
ghcr.io/microclaw/microclaw
What is microclaw?
MicroClaw is a Rust agent platform centered on one shared Agent Engine reachable through three interfaces. The Server operates as a persistent service handling chat integrations, web and API endpoints, scheduled jobs, and background automation. The Work application delivers a native GPUI desktop experience for local project workspaces with approval flows and checkpoints. The SDK enables Rust programs to embed the identical engine, skills, tooling, memory systems, and event runtime. Capabilities span multi-step tool chains, sessions that persist across restarts, guaranteed message delivery, file and SQLite storage with semantic recall, recurring and one-off tasks, and managed extensibility. It integrates with numerous chat services and web channels. The Server runs on macOS, Linux, and Windows; Work is production-ready on Apple Silicon macOS 13+ with preview builds elsewhere. Deployment options include a curl installer, Homebrew, Docker from ghcr.io, and source compilation. MIT licensed, currently at v0.10.0 (pre-1.0 beta).
Should you self-host microclaw?
microclaw scores 52/100 on selfhost.directory (how scoring works) and ranks #30 of 120 in AI Assistant. It has 749 GitHub stars.
It shipped 20 stable releases in the past year. The latest, v0.11.1, came out on 27 October 2026. New versions land roughly every 2 days.
Setup is moderate: expect some configuration beyond a single command. An official Docker image is available. It is MIT-licensed and written mostly in Rust.
Its closest alternative on selfhost.directory is hermes-agent (89/100, 257 926 stars). It scores higher overall, so compare the two before you pick. See microclaw vs hermes-agent side by side.
Features
- ClawHub skills
- Durable delivery
- Embeddable Rust SDK
- MCP tool federation
- Multi-step tool use
- Native desktop Work appOfficially supported on Apple Silicon macOS 13+; Linux/Windows portable previews
- Persistent memory
- Resumable sessions
- Scheduled Tasks
- Tool approvals
- Web UI
microclaw README
More in AI Automation
MindRoomSelf-hosted AI agent platform with persistent memory, 100+ tool integrations, and Matrix-based chat. Run locally or on your servers with local or cloud models per agent.
model-composemodel-compose is a self-hosted, open-source tool for building AI pipelines in YAML with local/cloud models, streaming, Gradio UI, and Docker deployment.
muxiMUXI is a self-hosted, open-source AI application server that lets you declare AI agents in YAML as deployable formations, then run them with a single command.
n8n-installSelfhost AI is an open-source Docker Compose installer for a complete AI homelab with n8n, Ollama, Open WebUI, AI agents, RAG, and monitoring behind automatic HTTPS.
nixopusNixopus is a self-hosted, open-source deployment platform with an AI agent that handles the full app lifecycle.
noraOpen-source control plane to run AI agent fleets on your infrastructure via Docker or Kubernetes with dashboard, API, CLI, and MCP automation.
Oh My SubagentsOh My Subagents is a self-hosted, open-source runtime that turns ad-hoc AI subagents into durable, accountable teams for Codex and Claude.
omadiaSelf-hosted, open-source agentic OS for running multi-agent AI teams on your own infrastructure with privacy controls and audit trails.
OmnaraOpen-source, self-hosted managed agent platform: run AI agents through one REST API with durable state, tool approvals, and any model or machine.
OmniOmni is a self-hosted, open-source AI agent connecting to workplace tools like Google Workspace, Microsoft 365, Slack, and Jira to gather context and execute tasks.
OmniParseSelf-hosted OmniParse converts documents, multimedia, and web content into structured markdown optimized for LLM apps like RAG. Open-source solution with Docker.
Open FlowOpen-source AI workflow automation platform where AI agents build workflows connecting 1,600+ apps, with visual Workbench for inspection and human approval.
0 comments