Security audit
zz-chat
Security checks for vulnerabilities and agentic risk
Overview
The skill mostly does what it claims (deploy a Cloudflare Worker and run a local bridge to OpenClaw) but includes persistent local services and communicates with third‑party servers (ai0000.cn and a GitHub Pages site) that receive your worker URL/ID and may see routed messages — this design is coherent for discovery but raises privacy/centralization concerns you should review before installing.
Before installing: 1) Review the bridge.py/bridge.js/watchdog.py sources yourself — they will be copied into ~/.openclaw and run as background processes. 2) Be aware the skill by default contacts external domains (https://ai0000.cn/zz/ and https://badxtdss.github.io) and embeds your Worker URL/ID in requests/QRs; if you don't trust those domains do not proceed. 3) If you want to minimize exposure, deploy the Cloudflare Worker to your account, ensure ~/.zz/worker_url contains your own worker URL, and start the bridge with --worker <your_worker_url> (this forces the bridge to use your worker instead of the default central server). 4) Avoid using the remote GitHub Pages UI: host the provided index.html/chat.html locally or on your own domain so your worker URL/ID aren't sent to a third party. 5) After installing, monitor outgoing connections (netstat/lsof) and inspect launchd/services to remove the auto-start if unwanted. 6) If you lack strong trust in the skill author or the central server, run this on an isolated machine or skip installation.
SkillSpector findings are pending for this release.