fact-check
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted external data (draft documents, source files, and web search results) while possessing high-privilege tool access.
- Ingestion points: Untrusted content is ingested in Phase 1 (EXTRACT) and Phase 2 (VERIFY) using the
ReadandWebFetchtools. - Boundary markers: The instructions lack explicit delimiters or instructions to the model to ignore potential commands embedded within the documents being verified.
- Capability inventory: The skill has access to sensitive tools including
Bash,Write, andWebFetch, which could be exploited if the model follows instructions contained within an attacker-controlled document. - Sanitization: No sanitization or safety-filtering of input data is defined before processing.
Audit Metadata