default-workflow

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external input such as task descriptions, codebase analysis, and human feedback, creating a surface for indirect prompt injection.
  • Ingestion points: task_description parameter, GitHub issue comments, PR review feedback, and local codebase analysis results.
  • Boundary markers: The skill relies on the recipe runner to enforce execution flow, but the provided documentation does not specify explicit delimiter-based prompt isolation.
  • Capability inventory: The skill has permissions to write files to the workspace, perform Git operations (commit, push), and interact with the GitHub API (create issues/PRs).
  • Sanitization: The documentation explicitly describes a resilience mechanism in Step 21 to prevent shell metacharacter injection by using heredocs and truncation for large template variables.
  • [SAFE]: The skill uses standard development tools (git, gh) for their intended purpose within a documented software development lifecycle.
  • [SAFE]: No obfuscation, remote code execution from unknown sources, or privilege escalation patterns were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 03:30 PM