skills/quantipixels/skills/atunwo/Gen Agent Trust Hub

atunwo

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill is composed of instructional markdown documents and configuration metadata. It lacks executable code, shell scripts, or patterns that trigger dangerous operations. The instructions emphasize maintaining a read-only state and restricting actions to analysis rather than execution or deployment.
  • [EXTERNAL_DOWNLOADS]: The skill references the official Rust documentation (doc.rust-lang.org) within the 'native-boundaries' guide to provide technical context for reviewing memory safety and foreign function interface (FFI) boundaries.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to evaluate external code changes, which creates an ingestion surface for untrusted data. The instructions mitigate risks through several architectural safeguards:
  • Ingestion points: Code snapshots and change sets provided by the user (SKILL.md).
  • Boundary markers: Explicitly mandates keeping source and Git state read-only and judging boundaries independently of provider-hosted content.
  • Capability inventory: The skill is restricted to inspection-only and does not authorize corrections, system mutations, or the execution of untrusted builds.
  • Sanitization: Instructions require resolving actual declarations and identities rather than trusting syntax matches or tool outputs blindly.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 08:14 AM
Security Audit — agent-trust-hub — atunwo