fact-check
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill establishes a logic-based verification workflow and does not contain any malicious code, external downloads, or attempts to access sensitive system files.
- [PROMPT_INJECTION]: The skill is designed to process untrusted external drafts, which constitutes an indirect prompt injection surface. However, the instructions include strong defensive measures by requiring the agent to maintain a skeptical posture, place the burden of proof on the speaker, and independently verify claims using lateral reading and primary sources.
- Ingestion points: Processes user-provided draft text and optional sender context (SKILL.md).
- Boundary markers: Absent for input data.
- Capability inventory: Implicit network search and URL access capabilities used for verification.
- Sanitization: No explicit sanitization of input text is defined, but the verification methodology requires independent corroboration, which mitigates the risk of following malicious instructions embedded in the draft.
Audit Metadata