skills/juliusbrussee/cavekit/check/Gen Agent Trust Hub

check

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external, untrusted content from the local workspace, which creates a potential surface for indirect prompt injection.
  • Ingestion points: Reads SPEC.md and various source files discovered during implementation checks (e.g., .go files mentioned in the documentation).
  • Boundary markers: The instructions do not define specific delimiters or "ignore instructions" wrappers for the content read from SPEC.md or source files.
  • Capability inventory: Limited to local file reading (read, grep). The skill explicitly states a "Zero writes" non-goal and does not invoke external tools or network requests.
  • Sanitization: There is no evidence of sanitization or validation of the content read from the files before it is processed by the agent's logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 05:17 PM
Security Audit — agent-trust-hub — check