rfp-ingest
Pass
Audited by Gen Agent Trust Hub on Feb 26, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: Indirect prompt injection surface identified. Untrusted data from external sources is ingested and stored without sufficient isolation. \n
- Ingestion points: RFP data is fetched from the SAM.gov API in
convex/ingestion.tsand parsed from user-uploaded CSV files inconvex/ingestion/rfpmartCsv.ts. \n - Boundary markers: Absent. The skill does not wrap the ingested RFP titles, descriptions, or summaries in delimiters to distinguish data from instructions. \n
- Capability inventory: The skill uses
ctx.runMutationto write ingested data to the database andfetchto communicate with external APIs. \n - Sanitization: Absent. No escaping, validation, or filtering of the external content (e.g., HTML or markdown in descriptions) is performed before storage. \n- [EXTERNAL_DOWNLOADS]: Fetches RFP opportunity data from the official SAM.gov government API. This is a well-known service and the connection is managed using environment variables for authentication.
Audit Metadata