research-add-fields
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data which could potentially contain malicious instructions.
- Ingestion points: Reads local data from
*/fields.yaml(Step 1) and ingests external content from theWebSearchtool (Step 2). - Boundary markers: No explicit delimiter or instruction-isolation markers are defined in the workflow to separate data from instructions.
- Capability inventory: The skill has access to
Bash,Read,Write, andWebSearchtools, allowing it to modify the local file system. - Sanitization: The skill includes a human-in-the-loop validation step where the user must confirm suggested fields before they are appended to the file (Step 3).
Audit Metadata