agui-dotnet-code-review
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions require the agent to execute local shell commands for repository and code analysis.
- Step 1 in SKILL.md directs the agent to determine the change set using
git merge-base <base> HEADto identify the files under review. - Rule
NET-ARCH-01in references/rules.md instructs the agent to rungit grep "Microsoft.AspNetCore" -- sdks/dotnet/srcto verify architectural boundaries.
- Step 1 in SKILL.md directs the agent to determine the change set using
- [INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted external data such as pull request bodies, branch content, and code diffs, which creates a surface for indirect prompt injection.
- Ingestion points: The agent processes external input in SKILL.md (Step 1) by reading specific pull request numbers, branches, or commit ranges to generate a diff for review.
- Boundary markers: The skill provides a structured template for the final 'Review summary' in SKILL.md, but does not specify delimiters or instructions to ignore potential commands embedded within the code diffs themselves.
- Capability inventory: The agent has the ability to read the local filesystem and execute
gitcommands as part of its analysis process (identified in Step 1 and NET-ARCH-01). - Sanitization: The instructions do not include specific sanitization, filtering, or escaping steps for the code content before it is processed by the agent or included in the review summary output.
Audit Metadata