skills/citypaul/dotfiles/find-gaps/Gen Agent Trust Hub

find-gaps

Pass

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data (plans, specs, and mocks) provided by the user. This creates a standard indirect prompt injection surface where instructions embedded in the artifacts could potentially influence the agent's behavior. However, the skill mitigates this risk through a high-friction interactive loop that requires explicit user confirmation before any modifications are written back to the source files.
  • Ingestion points: Reads user-provided artifacts including stories, plans, acceptance criteria, and design mocks.
  • Boundary markers: The skill does not define explicit delimiters for artifact content but operates on a structured checklist-based analysis.
  • Capability inventory: Includes the ability to write confirmed updates back to the source of truth (files, canvases, or docs).
  • Sanitization: No explicit sanitization of input text is performed, relying on the user's role in the 'Ask, capture, write back, confirm' loop.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 6, 2026, 08:26 AM
Security Audit — agent-trust-hub — find-gaps