The fastest way to run your own AI agents
A lightweight AI agent runtime in a single Go binary. Create, manage, and run AI agents — each with its own personality, memory, skills, and tools.
$ curl -fsSL https://raw.githubusercontent.com/fastclaw-ai/fastclaw/main/install.sh | bashDrops a single binary into ~/.local/bin. No dependencies.
Everything an agent needs, built in
FastClaw handles LLM communication, tool execution, sandbox isolation, and session management — so you can focus on your agents.
Agent Factory
Create and run many agents, each with its own SOUL.md personality, memory, skills, and tools. Per-agent model overrides and public chat sharing.
Any LLM provider
OpenAI, Anthropic, Ollama, OpenRouter, Groq, DeepSeek, Mistral, or any OpenAI-compatible API. Per-agent overrides and prompt caching.
Connect any channel
Bind Telegram, Discord, and Slack bots per agent. Sessions stay isolated per channel and chat, so threads never bleed into each other.
Tools & sandbox
Built-in exec, file, and web tools. Run them in an E2B cloud sandbox or Docker, with automatic skill + workspace hydration and post-exec sync.
Skills & plugins
Install shared skills from ClawHub or GitHub, agent-private or global. MCP server support and a JSON-RPC plugin system.
Memory & OpenAI-compatible API
Long-term MEMORY.md auto-updated by heartbeat, full session history, and a drop-in OpenAI-compatible /v1/chat/completions endpoint.
Running in under a minute
Install the binary, run the setup wizard, and open the dashboard. No config files — bootstrap is env-only.
- 1
Install
One command drops the fastclaw binary into ~/.local/bin and adds it to your PATH.
$ curl -fsSL https://fastclaw.ai/install.sh | bash - 2
First run
Launch fastclaw to open the setup wizard — configure an LLM provider and it creates your first agent.
$ fastclaw - 3
Open the dashboard
Manage agents, skills, models, channels, and API keys at http://localhost:18953.
$ open http://localhost:18953
Run it anywhere
From your laptop to a Kubernetes cluster — the same single binary, with the database as the only source of truth.
Local
Run in the foreground or install as a launchd / systemd daemon. SQLite by default — zero external infra.
Docker
Ship the prebuilt image with deploy/docker/start.sh. Env-only bootstrap, no config file to mount.
Kubernetes
Scale to multiple pods with Postgres storage and an S3-compatible object store for skills and files.
SQLite or Postgres
Default to a single SQLite file, or point FASTCLAW_STORAGE_DSN at Postgres for multi-pod deployments.
Source-available, yours to run
Use FastClaw commercially as a backend for your own product, or deploy it internally — free, under the FastClaw Community License. Building a multi-tenant SaaS on top? Reach out for a commercial license.
Frequently asked questions
Everything you need to know about running FastClaw.
Spin up your first agent today
One command to install, one to run. FastClaw is free to self-host and source-available — your agents, your infrastructure.
$ curl -fsSL https://raw.githubusercontent.com/fastclaw-ai/fastclaw/main/install.sh | bash