experience-completeness

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided journey signals and UI code to determine appropriate interface states (e.g., loading, error, success). This creates a surface where malformed or malicious signals could attempt to influence agent output.
  • Ingestion points: Journey signals and UI context are ingested from the active application flow to map concrete signals to states as defined in SKILL.md.
  • Boundary markers: The instructions do not specify the use of delimiters or 'ignore' commands when processing external data signals.
  • Capability inventory: The skill invokes several tools to manage and verify states, specifically safe-change, verify-before-done, and a2a-handoff.
  • Sanitization: The skill includes defensive guidelines in references/state-guide.md advising the agent to 'describe the actionable failure without leaking sensitive internals', which helps mitigate data exposure during error handling.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 05:53 AM
Security Audit — agent-trust-hub — experience-completeness