experience-selector

Pass

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the Job Description (JD) input. An attacker could embed malicious instructions within a JD to influence the agent's behavior or data selection process.
  • Ingestion points: The skill accepts a Job Description (JD) as a primary input for analysis in SKILL.md.
  • Boundary markers: Absent. There are no delimiters or instructions to ignore embedded commands within the JD content.
  • Capability inventory: The skill has the capability to read multiple sensitive local files (../../context/information.md, experience.md, etc.) to gather personal facts.
  • Sanitization: Absent. The skill does not describe any validation or filtering for the external JD input.
  • [DATA_EXPOSURE]: The skill accesses sensitive personal identifiable information (PII) stored in local files, which is necessary for its stated purpose but represents a data exposure surface.
  • Evidence: The instructions explicitly mandate reading ../../context/information.md (containing basic info, location, and phone numbers) and other professional history files.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 28, 2026, 09:50 AM