nx-cloud-ci-failure-review

Pass

Audited by Gen Agent Trust Hub on Oct 8, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external CI logs and pipeline metadata which could contain malicious instructions aimed at the agent.
  • Ingestion points: As described in SKILL.md under 'Scope and collection', the skill fetches records for CIPEs, run-groups, runs, terminal task rows, and representative task or instance log excerpts.
  • Boundary markers: The instructions do not define explicit delimiters or boundary markers to isolate the fetched external log content from the agent's core instructions.
  • Capability inventory: Although the skill body identifies the workflow as read-only, it operates within an agent context that likely possesses broader capabilities such as file system access or shell execution.
  • Sanitization: The skill provides instructions to 'Quote only redacted error lines' in the final report, which serves as a basic output filter, but it does not specify sanitization or escaping of external content during the ingestion and reasoning phases.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 8, 2026, 06:05 PM