cross-provider-translation

Warn

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill automates the installation of Git hooks, which are executable scripts that run automatically during version control operations.
  • Evidence: The 'Bootstrap Mode' and 'Convention Validation' sections instruct the agent to install 'commit-msg' and 'pre-push' hooks into the .git/hooks/ or .husky/ directories.
  • [REMOTE_CODE_EXECUTION]: The skill automates the deployment of CI/CD configurations, which can execute arbitrary code in automated build environments.
  • Evidence: The 'Bootstrap Mode' section (Step 5) requires copying GitHub Actions workflows to the .github/workflows/ directory.
  • [PROMPT_INJECTION]: The skill functions as a multi-step indirect prompt injection surface by ingesting, reformatting, and deploying external skill files from the docs/members/ directory.
  • Ingestion points: Reads all member files located at docs/members/the-<name>/SKILL.md (SKILL.md).
  • Boundary markers: None identified; the skill directly processes and translates the content of these files.
  • Capability inventory: The skill performs file system writes (translation output), Git hook installation, and CI workflow creation across all referenced provider targets (SKILL.md).
  • Sanitization: No evidence of content validation or escaping is present; the skill assumes the source skill files are trustworthy.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 8, 2026, 10:05 AM
Security Audit — agent-trust-hub — cross-provider-translation