github-workflow-automation
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill outlines workflows that ingest and process external untrusted data, introducing an indirect prompt injection attack surface.
- Ingestion points: Fetches external pull request details via
gh pr viewwithinSKILL.mdworkflow examples. - Boundary markers: Absent. No clear delimiters or formatting boundaries are implemented to isolate untrusted pull request data from automated commands.
- Capability inventory: Invokes
npx ruv-swarm actions pr-validatewith the fetched pull request data passed directly into the tool execution. - Sanitization: Absent. No validation or character filtering is applied to the incoming repository data prior to interpretation.
- [EXTERNAL_DOWNLOADS]: The skill demonstrates execution of external CLI tools and modules via
npx ruv-swarmand instructions to installclaude-flow@alpha. These components align with the author context and the declared execution prerequisites.
Audit Metadata