agentsight-code-review
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands to retrieve project history and code changes for the review process.
- Evidence: Step 1 in
SKILL.mdexecutesgit diff origin/main..HEAD,git diff --stat origin/main..HEAD, andgit log --oneline origin/main..HEAD. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from source code changes. Malicious instructions hidden in code comments or files could influence the agent's logic during the review.
- Ingestion points: The output of the
git diffcommand is directly processed by the agent inSKILL.md. - Boundary markers: None; the skill lacks specific instructions to ignore or delimit embedded commands within the diff content.
- Capability inventory: The skill has access to shell execution via
gitcommands. - Sanitization: No sanitization or verification is applied to the code content before analysis.
Audit Metadata