code-review
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from git diffs, commit messages, issue trackers, and project documentation. This creates a surface for indirect prompt injection where malicious content within the reviewed files could attempt to influence the reviewer's output.
- Ingestion points: Git diff output, git logs, GitHub/GitLab CLI data (
gh), and local planning/documentation files. - Boundary markers: The skill structures sub-agent prompts with specific briefs but does not include explicit delimiters or instructions to ignore embedded commands within the ingested code or spec content.
- Capability inventory: The skill utilizes shell commands (
git,gh) and spawns parallel sub-agents via the platform'sAgenttool. - Sanitization: No explicit sanitization or escaping of the ingested diff or documentation content was identified.
- [COMMAND_EXECUTION]: The skill executes local shell commands using
gitandghto resolve references and fetch diffs. These operations are restricted to standard repository metadata retrieval and are consistent with the skill's primary purpose. - [SAFE]: The skill contains 'Mandatory risk checks' which proactively instruct the agent to identify security vulnerabilities in the code under review, such as credential leaks, unsafe data handling, and improper boundary checks. This is a positive security practice.
Audit Metadata