codeguard-metal
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill follows security best practices by explicitly instructing the agent not to install dependencies over the network, not to record sensitive credentials or tokens, and to avoid hallucinating commands when tools are missing.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface through the processing of untrusted repository configurations and CI commands, but provides sufficient defensive measures to mitigate these risks.
- Ingestion points: Local repository configuration files and CI command definitions are read to align diagnostic checks (SKILL.md, Step 2).
- Boundary markers: The skill requires the agent to explicitly report status as 'UNVERIFIED' or 'PLANNED' when tools are unavailable or configurations are unclear, preventing the agent from being coerced into executing unauthorized logic (SKILL.md, Execution Contract).
- Capability inventory: Shell command execution is limited to tool detection and diagnostic checks, and any file modifications require mandatory diff review by the user (SKILL.md, Gotchas).
- Sanitization: The skill includes warnings against using literal placeholders and mandates the use of authorized file paths for all operations (SKILL.md, Gotchas).
Audit Metadata