business-operations-skills

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local working directory to perform routing and operational analysis. This data ingestion creates a surface where instructions hidden inside files (like vendor catalogs or process transcripts) could influence the agent's actions or outputs.
  • Ingestion points: The skill uses grep to explore the working directory for files including vendor catalogs, process transcripts, org charts, and CSV files (e.g., procurement-Q3.csv) as described in SKILL.md.
  • Boundary markers: The instructions do not specify any delimiters or warnings to ignore instructions embedded within the processed data.
  • Capability inventory: The skill can write artifacts (markdown, CSV, JSON) to the working directory and execute Python tools using the standard library.
  • Sanitization: No sanitization, filtering, or validation steps are mentioned for the content read from external files.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 04:56 PM
Security Audit — agent-trust-hub — business-operations-skills