1. Install
Use the virtualenv quickstart or installer so Libre Claw stays isolated from system Python.
Libre Claw documentation
Browse Libre Claw docs: install the CLI, configure model providers, launch the dashboard, connect Libre WebUI, wire Telegram, schedule runs, use skills, and audit security.
Install Libre Claw
Install Libre Claw in a Python virtual environment, launch the TUI, and start the daemon or Telegram bridge.
git clone -> venv -> libre-claw tui Docs 02 Provider and model routing
Configure Moonshot AI, OpenRouter, Ollama Cloud, Codex OAuth, Anthropic, OpenAI, local Ollama, and fallback model routes in Libre Claw.
/model openrouter:... --global Docs 03 Local dashboard
Use the Libre Claw dashboard to inspect runs, approve blocked tools, edit schedules, trigger jobs, and review token usage.
http://127.0.0.1:8766/dashboard Docs 04 Libre WebUI integration
Connect Libre WebUI to the Libre Claw daemon for a browser control surface with run timelines, inline tool approvals, automations, and provider usage.
libre-claw start Docs 05 Telegram setup
Connect Libre Claw to Telegram, authorize numeric user IDs, approve tools inline, and receive scheduled reports.
libre-claw telegram setup Docs 06 Schedules and heartbeats
Create recurring Libre Claw runs, route reports to Telegram or files, trigger jobs immediately, and keep periodic heartbeat checks.
/schedule add ... Docs 07 Skills, Vercel Skills, and SOUL.md
Use local Markdown skills, optional Vercel Skills discovery, and SOUL.md persona files to make Libre Claw reuse project workflows safely.
/skills sync Docs 08 Security and privacy
Review Libre Claw's key storage, approval policy, sandboxing, memory redaction, daemon locality, and audit trail.
/approvals 2. Choose a provider
Save a model globally, add fallbacks, and keep provider keys out of project files.
3. Run and review
Start from the TUI, daemon, dashboard, or Telegram. Approve side effects and inspect artifacts later.