ipd-skill
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- File System Write Consideration: The
scripts/rest_request.pyscript includes a feature to save raw API data to a path specified by theraw_output_pathparameter. This allows for writing content to arbitrary directories where the agent has permissions, which could be used as a vector for persistence or unauthorized file modification if the path is not strictly controlled. - External Data Processing Surface: The skill ingests and processes content from an external REST API, creating a potential surface for indirect prompt injection. Since the skill also has file-write and network capabilities, there is a possibility that instructions embedded in the external data could attempt to trigger unintended actions. Ingestion points: Data is fetched from a remote endpoint in
scripts/rest_request.pyusing therequestslibrary. Boundary markers: No explicit delimiting markers or instructions to ignore embedded commands are used when presenting data to the agent. Capability inventory: Thescripts/rest_request.pyscript performs network operations and file system writes. Sanitization: The script performs basic string truncation for display purposes but does not sanitize or filter the content of the API response for potentially malicious instructions.
Audit Metadata