Ai Poison Guard

• Verdict: Pass
• Analyzed: May 11, 2026, 6:57 AM
• Findings: 0
• Version: 1.0.0
• Hash: 9f9e3c9603627409cde7f272aad00d779e9e593b061ee48886c36d2869551d65
• Source: palm
• External report: View on VirusTotal

Overview

Type: OpenClaw Skill Name: ai-poison-guard Version: 1.0.0 The skill is a text-scanning utility designed to detect 'AI poisoning' or GEO (Generative Engine Optimization) patterns using regex-based matching. The primary script, `scripts/guard.py`, implements simple keyword and pattern detection for marketing-speak and 'fake authority' claims without any network activity, data exfiltration, or suspicious execution logic. While the documentation in `SKILL.md` mentions advanced features like URL verification and domain reputation that are not yet implemented in the code, the existing functionality is harmless and aligns with the stated security-focused purpose.

Findings (0)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

ASI09: Human-Agent Trust Exploitation

NoteHigh Confidence

What this means

The tool may give users more confidence than warranted, especially for website/source verification that the included code does not perform.

Why it was flagged

The shipped script only supports local text/file detection and JSON output, while SKILL.md advertises source verification, domain checks, URL detection, and scoring commands. This is not malicious, but users should not rely on unimplemented security claims.

Skill content

parser.add_argument("--detect-text"...); parser.add_argument("--detect-file"...); parser.add_argument("--json"...)

Recommendation

Treat this as a simple heuristic content scanner unless the missing verification features are actually implemented and reviewed.

ASI04: Agentic Supply Chain Vulnerabilities

NoteHigh Confidence

What this means

Installing unnecessary unpinned dependencies can add avoidable supply-chain exposure, although nothing in the artifacts runs this automatically.

Why it was flagged

The documentation suggests a manual, unpinned package install, but requirements.txt says the current version is pure local detection and the script does not import these packages.

Skill content

pip3 install requests whois tldextract

Recommendation

Do not install the optional packages unless needed; if future versions require them, prefer pinned versions and reviewed dependency metadata.