agentic-workflows

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches extensive instructional and configuration content from the github/gh-aw repository. These files are used to direct the agent's behavior during workflow design, debugging, and maintenance tasks.
  • [INDIRECT_PROMPT_INJECTION]: The skill allows for repository-specific instruction overrides through a local configuration file, creating a surface for indirect injection if the repository is untrusted.
  • Ingestion points: Loads local repository files (e.g., .github/aw/instructions.md) and remote repository content (github/gh-aw).
  • Boundary markers: Absent; instructions from external files are loaded and followed without explicit delimiters or warnings to ignore embedded instructions.
  • Capability inventory: High; the agent is tasked with designing, creating, debugging, and patching workflows, which typically implies file system modifications and execution of CLI tools.
  • Sanitization: Absent; content from the loaded files is designed to be followed directly by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 04:42 AM