vendor-manager

Pass

Audited by Gen Agent Trust Hub on Apr 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill contains no patterns of prompt injection, obfuscation, or malicious command execution. Its operations are limited to processing business information and generating markdown-based reports.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (vendor profiles and project details), which is a common surface for indirect prompt injection.
  • Ingestion points: User-provided vendor data and project scope definitions processed through the risk management workflow in references/risk-register.md.
  • Boundary markers: The instructions do not implement specific delimiters or system-level instructions to ignore embedded commands within the vendor data.
  • Capability inventory: The skill utilizes the mcp__sequential-thinking__sequentialthinking tool for analytical tasks and performs file-write operations to a local _workspace/ directory to organize output.
  • Sanitization: No explicit data sanitization or validation logic is present, though the impact is limited as the output is primarily descriptive markdown.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 13, 2026, 11:42 PM