Back to skill
Skillv1.0.0
VirusTotal security
AgentWard Sanitize · External malware reputation and Code Insight signals for this exact artifact hash.
Scanner verdict
BenignApr 30, 2026, 4:35 AM
- Hash
- 76c65e48319b55d7b75b05648fab764bebdb8f1e169a87971ece4b5c6f093d1e
- Source
- palm
- Verdict
- benign
- Code Insight
- Type: OpenClaw Skill Name: sanitize Version: 1.0.0 The skill is designed for PII detection and redaction, with strong explicit safeguards against data leakage, especially when used by an AI agent. The `SKILL.md` contains clear instructions to the agent to 'Do NOT read the input file directly' and 'Do NOT read the entity map file', which are effective prompt injection defenses. The `scripts/sanitize.py` code confirms zero external dependencies and no network calls, and it meticulously prevents raw PII from being output to stdout in `--json` or `--preview` modes, instead writing sensitive mappings to a separate, warned-about sidecar file on disk. All observed behaviors are aligned with the stated purpose and demonstrate a clear intent for secure operation.
- External report
- View on VirusTotal