devops-infrastructure-engineer
Warn
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions reference and utilize multiple third-party external agent modules for automated testing and auditing, specifically
stablyai/orca@computer-use,midscene-skills@computer-automation, andcodex-skills@gemini-computer-use. These external dependencies are not from recognized trusted sources and could introduce security risks if the referenced modules are updated with malicious code or perform unauthorized actions when invoked. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill reads project-specific infrastructure and tech stack requirements from the
.agents/startup_dna.mdfile. - Boundary markers: There are no defined delimiters or safety instructions (e.g., "ignore embedded instructions") specified for handling the content ingested from the project configuration file.
- Capability inventory: The skill is authorized to create and modify critical deployment infrastructure, including CI/CD YAML workflows, Docker configurations, and automated backup/recovery scripts triggered by cron.
- Sanitization: The instructions lack any validation or sanitization steps to ensure that the ingested tech stack specifications do not contain malicious instructions that could influence the generated infrastructure code.
Audit Metadata