A control panel for your AI chatbot army

What it does

ClawPanel is a web-based management panel for OpenClaw, an AI assistant framework. It wraps installation, configuration, process control, and monitoring into one deployable Go binary with an embedded React frontend. The panel speaks to external runtimes—OpenClaw gateway, NapCat for QQ, WeChat bridges, and roughly twenty other messaging platforms—rather than bundling them, so upgrades stay independent.

The interesting bit

The project treats “AI agent infrastructure” like traditional server admin. There’s a dashboard for gateway health, a topology graph showing how channels route to agents, a workflow canvas with nodes for ai_plan and approval, and even an “AI Company” feature that assigns tasks to virtual roles with review cycles. It’s essentially systems management vocabulary applied to conversational bots.

Key highlights

• Single Go binary with go:embed frontend; deploys via systemd, launchd, or Windows Service
• Manages 20+ channels: QQ, WeChat, Feishu, DingTalk, Telegram, Discord, WhatsApp, Slack, Matrix, LINE, Twitch, and others
• Real-time log streaming over WebSocket; SQLite persistence for events, sessions, and tasks
• Visual workflow builder with human-in-the-loop steps: approval, wait-for-user, retry, file handoff
• Built-in updater with SHA256 verification and binary rollback; pulls from GitHub Releases only
• Configuration manager auto-cleans legacy openclaw.json fields to maintain schema compatibility

Caveats

• CC BY-NC-SA 4.0 license: explicitly non-commercial, with a stern disclaimer about Tencent ToS risks for QQ/WeChat automation
• Lite distribution line was killed in v5.5.1; only “standard” edition remains, which does not include OpenClaw runtime
• Some platform integrations (NapCat Docker vs. Windows Shell, WSL bridging) have documented rough edges

Verdict

Worth a look if you’re already running OpenClaw or managing multiple bot channels and want a unified control plane. Skip it if you need a fully open license, a batteries-included distribution, or a simple single-channel bot—this is ops tooling for multi-agent deployments, not a chatbot framework itself.