ai-impact-assessment

Pass

Audited by Gen Agent Trust Hub on Mar 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of untrusted user data without sufficient isolation or sanitization.\n
  • Ingestion points: The skill reads and analyzes user conversation history, CV content, and previous output files from the current working directory to inform its research phase (as seen in references/ai-impact-research.md).\n
  • Boundary markers: There are no explicit delimiters or specific prompt instructions to isolate untrusted user data from the agent's core task instructions.\n
  • Capability inventory: The skill possesses capabilities to perform external web searches and create new markdown files on the local filesystem (documented in SKILL.md and references/assessment-and-plan.md).\n
  • Sanitization: The implementation lacks mechanisms to sanitize or validate user-provided data before it is interpolated into search queries or the final assessment report.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 17, 2026, 09:25 AM
Security Audit — agent-trust-hub — ai-impact-assessment