ci-monitoring

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • PROMPT_INJECTION (LOW): The skill is susceptible to Indirect Prompt Injection through untrusted GitHub pull request data.
  • Ingestion points: SKILL.md (reading review comments via gh api graphql) and reference/ci-failures.md (reading CI logs via gh run view).
  • Boundary markers: Absent. No delimiters or specific 'ignore instructions' warnings are provided for external data processing.
  • Capability inventory: Access to Bash (shell execution), Edit (file modification), and mcp__github__* (GitHub management).
  • Sanitization: Absent. The skill relies on the LLM's 'understanding' and 'verification' of feedback rather than technical filtering.
  • COMMAND_EXECUTION (LOW): The skill executes local environment scripts (pnpm test, pnpm build, pnpm lint) which are defined within the repository being monitored.
  • Evidence: SKILL.md and reference/ci-failures.md contain instructions to run local build and test suites to reproduce CI failures.
  • Risk: If an attacker submits a malicious PR that modifies package.json scripts or test files, the autonomous nature of the skill ('Do NOT stop to report or ask') may lead to the execution of malicious code in the local environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:13 PM