devops-plan
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the Bash tool in SKILL.md to execute file discovery commands (ls) and content searching (grep). These operations are used for the legitimate purpose of detecting existing Infrastructure-as-Code (IaC) files and cloud providers.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from local files, including Terraform (.tf), Ansible, and Cloud-init configurations, to generate a structured context. This constitutes a surface for indirect prompt injection as it ingests untrusted local data to influence agent recommendations.
- Ingestion points: Local file system via ls and grep on IaC configuration files in SKILL.md (e.g., *.tf, ansible.cfg, playbook.yml).
- Boundary markers: Absent; data is extracted and listed directly without explicit delimiters or safety instructions regarding embedded content.
- Capability inventory: SKILL.md performs Bash execution (ls, grep, tr, sort) and invokes external skills (infra-security-review).
- Sanitization: Absent; provider names and tool presence are extracted as raw strings from file content.
Audit Metadata