comp-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill is designed to process external datasets, such as CSV files or pasted compensation bands, to perform benchmarking and outlier analysis. While this is necessary for its core functionality, it creates a potential surface where untrusted data could influence agent behavior.
- Ingestion points: The skill explicitly requests users to "Upload a CSV or paste your comp bands" in SKILL.md.
- Boundary markers: There are no explicit delimiters or instructions defined in the skill to isolate the ingested data from the agent's control flow, which could allow instructions embedded in a CSV to be interpreted as commands.
- Capability inventory: The skill utilizes internal connectors (~~compensation data, ~~HRIS) to fetch and compare sensitive market and employee data.
- Sanitization: The skill does not describe specific validation or sanitization routines for the uploaded data, relying instead on the agent's default processing logic.
Audit Metadata