auto-skill-lifecycle-handling

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to ingest untrusted data from user feedback and convert it into persistent instructions (Indirect Prompt Injection surface).
  • Ingestion points: The skill monitors user messages and feedback to identify "stable constraints" and "preferences" for skill extraction (found in SKILL.md prompt section).
  • Boundary markers: There are no instructions or mechanisms defined to use delimiters or ignore embedded instructions when the extracted content is injected into future prompt contexts.
  • Capability inventory: The skill has the capability to generate new SKILL.md files and inject their contents into the LLM's context during subsequent queries.
  • Sanitization: The workflow lacks any sanitization, filtering, or validation steps to ensure that extracted constraints do not contain malicious payloads intended to hijack the agent's behavior in future sessions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 11:17 AM
Security Audit — agent-trust-hub — auto-skill-lifecycle-handling