audit-realworld

Pass

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill functions as a read-only auditor for architectural and functional gap analysis. It is explicitly designed to report findings without modifying the codebase.
  • [EXTERNAL_DOWNLOADS]: The skill fetches API specifications and test suites from the official RealWorld GitHub repository (realworld-apps/realworld) and references the hosted benchmark API (api.realworld.show). These are established, well-known resources within the developer community for benchmarking full-stack implementations.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it processes untrusted data from the repository being audited. 1. Ingestion points: Dependency manifests, source code, and test suite files (Phase 0, 1, 4). 2. Boundary markers: Absent; instructions do not specify how the agent should distinguish between repo content and instructions. 3. Capability inventory: The agent is instructed to execute command-line tools such as bru, hurl, newman, and playwright-cli (Phase 4) against the local environment. 4. Sanitization: Absent. While this surface exists, it is inherent to the auditing task performed by the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 7, 2026, 11:26 PM