github-actions-creator
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze project files (such as
package.json,requirements.txt, andREADME.md) to identify the technology stack. This is a potential surface for indirect injection if project files contain malicious instructions. - Evidence: Step 1 in
SKILL.mddefines the ingestion of project metadata. - Mitigation: The skill explicitly includes a 'Security Best Practices' section that directs the agent to prevent script injection in the generated output by using environment variables instead of direct event interpolation (e.g., using
env: ISSUE_TITLE: ${{ github.event.issue.title }}instead ofrun: echo "${{ github.event.issue.title }}"). - [EXTERNAL_DOWNLOADS]: The workflow templates reference official and well-known GitHub Actions for setup, building, and deployment.
- Evidence: References include actions from
actions/(GitHub),docker/,aws-actions/,google-github-actions/,cloudflare/, andhashicorp/. - Status: These are recognized as trusted or well-known services and are used for their intended purpose in CI/CD automation.
- [COMMAND_EXECUTION]: The templates provide standard build and test commands for various ecosystems.
- Evidence: Includes commands like
npm ci,pip install,go test,cargo clippy, andterraform apply. - Status: These are routine operations within a development workflow.
Audit Metadata