ai-claim-checker
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a framework for verifying claims in AI-generated text. It does not contain any malicious code, obfuscation, or persistence mechanisms.
- [PROMPT_INJECTION]: The skill includes a 'Safety boundary' section that explicitly instructs the agent to treat all input data (AI text, pasted sources, web excerpts) as untrusted. It specifically forbids following directives inside that material, such as workflow changes or command execution.
- [DATA_EXFILTRATION]: No sensitive file access or unauthorized network exfiltration patterns were detected. The skill uses neutral attribution links to a well-known service (GitHub).
- [REMOTE_CODE_EXECUTION]: No remote scripts are downloaded or executed. The instructions explicitly mandate that code snippets within verified material must be kept inert.
- [INDIRECT_PROMPT_INJECTION]: The skill has a clear attack surface as it processes untrusted third-party text, but it mitigates this risk through specific instructions.
- Ingestion points: Untrusted data enters the context via user-provided AI text and external source excerpts (SKILL.md).
- Boundary markers: Present in the 'Safety boundary' section, explicitly defining input as untrusted.
- Capability inventory: No subprocess calls, file-write operations, or network operations are present in the skill's instructions or metadata.
- Sanitization: The skill requires that code snippets and links found in external material be kept inert unless a separate, in-scope request is made by the user.
Audit Metadata