Meeting Risk Radar

会前识别高风险议题、模糊责任、缺失材料和可能失控的讨论点。；use for meeting-risk, preflight, facilitation workflows；do not use for 分析私密录音, 替代正式风险审查.

This skill appears to do what it says: generate pre-meeting risk briefs from local input. Before installing or running it: (1) review the bundled scripts/run.py yourself (it’s included) and run smoke-test in a safe directory; (2) never point the script at system roots or directories containing secrets—only provide meeting materials; (3) prefer --dry-run first and inspect output before allowing any follow-up actions; (4) note the homepage/owner are not a known vendor—if provenance matters, run in a sandbox or ask the publisher for verification; (5) do not supply private recordings or other sensitive PII to the skill.

Type: OpenClaw Skill Name: meeting-risk-radar Version: 1.0.0 The skill bundle is classified as suspicious due to a significant discrepancy between its stated purpose (meeting facilitation and risk identification) and the actual capabilities of the included Python script. While the documentation focuses on meeting agendas, `scripts/run.py` contains undocumented functionality for scanning the filesystem for sensitive data, including regex patterns for API keys, tokens, and dangerous shell commands (e.g., `curl|bash`, `rm -rf`). Although the script includes logic to redact secrets and lacks explicit exfiltration code, the inclusion of a multi-mode security scanner and a 'Skill Audit' feature within a 'Meeting Risk Radar' tool is highly irregular and suggests potential for unauthorized reconnaissance or undocumented behavior.