grant-proposal-review
Pass
Audited by Gen Agent Trust Hub on Apr 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it processes untrusted data from user-uploaded grant application documents (PDF or text).
- Ingestion points: The skill reads application files via the
Readtool as specified in the required inputs. - Boundary markers: The instructions do not define clear delimiters or use "ignore embedded instructions" directives when processing the document content, making the agent potentially susceptible to instructions hidden within the grant proposal.
- Capability inventory: The agent has access to
WebSearch,WebFetch, andBashtools. If a malicious application file contained adversarial instructions or payloads, the agent might inadvertently execute them using these tools. - Sanitization: There are no explicit content validation or sanitization steps for the text extracted from the uploaded files before it is processed by the agent.
Audit Metadata