Security audit

ClawWatch

Security checks for vulnerabilities and agentic risk

Overview

ClawWatch appears to do what it says: run a Node-based telemetry agent that reports OpenClaw/node metrics to a ClawWatch Worker.

Install this only if you intend to send node telemetry to ClawWatch, especially the default Worker at https://cw.osglab.win unless you configure a self-hosted Worker. Expect it to run continuously after the plugin is enabled and to report system metrics, token/status fields, and some node metadata such as local IP/region/GPU model. Protect ~/.clawwatch/agent.json because it contains the node secret. Confidence is medium rather than high because the provided src/agent.mjs content was truncated, so the full runtime agent code was not completely reviewable here.

Static analysis

Detected: suspicious.dangerous_exec, suspicious.env_credential_access

Shell command execution detected (child_process).

Critical

Code: suspicious.dangerous_exec
Location: dist/index.js:43
Evidence: runChild = spawn(process.execPath, [agentScript, "run", "--base", base], {

Shell command execution detected (child_process).

Critical

Code: suspicious.dangerous_exec
Location: src/agent.mjs:226
Evidence: const out = execSync('df -h / | tail -1', { timeout: 5000 }).toString().trim();

Shell command execution detected (child_process).

Critical

Code: suspicious.dangerous_exec
Location: src/index.ts:63
Evidence: runChild = spawn(process.execPath, [agentScript, 'run', '--base', base], {

Environment variable access combined with network send.

Critical

Code: suspicious.env_credential_access
Location: src/agent.mjs:43
Evidence: const LANG = (process.env.OPENCLAW_LANG || Intl.DateTimeFormat().resolvedOptions().locale || 'en')