define-security-policy
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Command Execution: The skill utilizes a Python script to resolve security policy chains within a repository. This is an intended function using repository paths and scopes as parameters. The process incorporates checks, such as verifying file sizes and types, to support stable operation.
- Indirect Prompt Injection: The skill processes repository files that could potentially contain untrusted instructions. It mitigates this consideration by instructing the agent to treat such content as evidence for review rather than as authorization for actions. 1. Ingestion points: The skill reads source code, configurations, and existing security policies (SKILL.md). 2. Boundary markers: It provides explicit instructions to treat repository data as untrusted and to display control characters visibly (SKILL.md). 3. Capability inventory: The skill involves executing a resolution script and the potential modification of security policy files (SKILL.md). 4. Sanitization: It implements a 1 MiB limit for file processing and requires visible character rendering in diff previews to prevent UI-based obfuscation.
Audit Metadata