improve-skill

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it ingests and processes untrusted skill definitions which could contain adversarial instructions.
  • Ingestion points: Target SKILL.md and directory contents loaded in Phase 0.
  • Boundary markers: The skill does not employ explicit delimiters or instructions to ignore embedded commands within the analyzed content.
  • Capability inventory: The agent has access to Edit, Write, WebFetch, and Task tools used throughout Phases 2, 4, and 5.
  • Sanitization: No sanitization or validation of the input skill's content is performed prior to processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 03:40 PM