docgen-generate
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill includes instructions that override the agent's standard behavioral boundaries regarding file processing, specifically directing it to process PDFs via a remote API instead of refusing the request.
- [EXTERNAL_DOWNLOADS]: User-provided document data and uploaded PDF files are transmitted to an external service (docgen-production-503d.up.railway.app) for generation and analysis. This service is hosted on Railway, a well-known platform, and represents the vendor's infrastructure.
- [COMMAND_EXECUTION]: The instructions require the agent to save binary PDF content received from the remote API directly to the local file system.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its data extraction workflow. 1. Ingestion points: Data is extracted from external PDF files via the /api/v1/scan endpoint. 2. Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the instructions for handling extracted data. 3. Capability inventory: The skill can perform network operations and local file writing. 4. Sanitization: No explicit validation or filtering of the extracted content is described before it is used to populate subsequent generation requests.
Audit Metadata