extract-tables
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted PDF files to extract table data and captions. If a malicious PDF contains text specifically designed to influence the agent's behavior during or after extraction, it could lead to indirect prompt injection.
- Ingestion points: Reads content (tables and captions) from user-provided PDF files in SKILL.md.
- Boundary markers: No specific delimiters or instructions to ignore embedded instructions are defined for the extraction process.
- Capability inventory: The skill performs file system write operations (creating directories, writing CSV and manifest files) and file reads.
- Sanitization: While the skill mentions sanitizing captions for filenames (slug generation), it does not specify sanitization for the content written into the CSV or manifest files.
Audit Metadata