fsi-ib-pitch-deck
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands (
sofficeandpdftoppm) within its validation loop to convert PowerPoint files to PDF and then to images. This is a legitimate functional requirement for visual verification of slide rendering as described in the workflow. - [INDIRECT_PROMPT_INJECTION]: The skill defines a high attack surface for indirect prompt injection due to its core function of ingesting and processing untrusted data from multiple sources (Excel, CSV, PDF, Word, and web sources). \n
- Ingestion points: Multiple external file formats and web sources specified in 'Phase 1: Data Extraction'. \n
- Boundary markers: There are no explicit instructions for the agent to use delimiters or sanitization logic to isolate external data from its internal instruction context. \n
- Capability inventory: The agent has file system access for creating backups and outputs, and execution capabilities for document conversion tools. \n
- Sanitization: No explicit data sanitization or validation steps for input content are mentioned beyond numerical verification.
Audit Metadata