ClaudeClaw is a Claude Code plugin that turns a Claude Code installation into an always-on personal assistant. It runs as a background daemon that executes scheduled tasks, responds to messages on Telegram, Discord, and Slack, transcribes voice commands, and connects to other services, using your existing Claude Code subscription rather than a separate API integration.
You install it through the Claude Code plugin marketplace and run /claudeclaw:start; a setup wizard configures the model, heartbeat, messaging channels, and security level, after which the daemon starts with a web dashboard. Memory uses Claude Code's built-in memory system plus CLAUDE.md, and isolation is folder-based.
Access is allowlist-based: an empty allowedUserIds list blocks everyone, and the daemon refuses to start when a bot token is configured without at least one allowed user ID. All /api/* routes except /api/health require a bearer token that is generated on first start.
Features
- Heartbeat: periodic check-ins with configurable intervals, quiet hours, and editable prompts
- Cron jobs: timezone-aware schedules for repeating or one-time tasks
- Telegram: text, image, and voice message support
- Discord: DMs, server mentions and replies, slash commands, voice messages, and image attachments (text attachments truncate at 2,048 bytes)
- Slack: a Socket Mode bot handling DMs, channel mentions, threads, voice messages, and file attachments
- Discord thread sessions: each thread gets its own isolated Claude CLI session, runs concurrently with others, bootstraps on first message, and is cleaned up when the thread is deleted or archived
- Time awareness: message time prefixes give the agent context about delays and daily patterns
- GLM fallback: continues with GLM models when the primary model's usage limit is reached
- Web dashboard: manage jobs, monitor runs, and inspect logs in real time
- Security levels: four access levels from read-only to full system access
- Model selection: switch models per workload
