github-workflow-automation

Pass

Audited by Gen Agent Trust Hub on Oct 16, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data retrieved from GitHub (via gh pr view and gh run view) and from repository analysis (stack detection) to drive its automation logic. This creates a surface where an attacker could embed malicious instructions in a pull request body or codebase file to manipulate the swarm's coordination.
  • Ingestion points: Data consumed from Pull Request metadata, workflow run logs, and codebase stack analysis.
  • Boundary markers: The workflow examples do not demonstrate the use of explicit delimiters or specific 'ignore instructions' guards for external data within the orchestration process.
  • Capability inventory: The skill has access to the GitHub CLI for high-impact actions including creating issues, commenting on pull requests, and managing repository structure.
  • Sanitization: While the provided templates use jq and base64 for reliable data handling, no explicit filtering or validation of processed content for adversarial instructions is included.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 16, 2026, 03:08 PM