gh-aw-operations
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill contains templates for workflows that ingest external data via GitHub triggers.
- Ingestion points: Workflow patterns in Skill 7 ingest data from GitHub issues, issue comments, and pull requests.
- Boundary markers: The templates do not explicitly show delimiters or specific instructions to ignore embedded content, though the skill body discusses the 'safe-outputs' framework features to limit write impact.
- Capability inventory: The documented workflows across all skills use
bash,edit(file read/write), andsafe-outputs(which wrap GitHub API write operations like creating PRs or issues). - Sanitization: No explicit sanitization or filtering of the external content (like PR bodies or comments) is shown in the prompt templates.
- [EXTERNAL_DOWNLOADS]: The skill provides examples of using Model Context Protocol (MCP) servers and importing agent definitions from GitHub. These references target the author's repositories or well-known technology services and are documented neutrally as intended framework features.
- [COMMAND_EXECUTION]: The documentation includes standard usage patterns for the
gh awCLI tool to compile, validate, and manage workflow files.
Audit Metadata