code-to-review
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses git shell commands to analyze the repository state.
- Evidence:
git rev-parse,git diff,git log, andgit check-ignoreare called withinSKILL.mdto identify commits and extract changes for review. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from repository files that could potentially contain malicious instructions.
- Ingestion points: The agent reads file contents via
git diff, along with specification documents indocs/specs/, implementation plans indocs/plans/, and ADR files. - Boundary markers: The skill employs parallelized sub-agent briefs to isolate the analysis of standards from specifications, providing basic context separation.
- Capability inventory: The skill has the ability to execute git commands, write review reports to the filesystem, and append new tasks to implementation plans.
- Sanitization: No explicit filtering or sanitization of ingested code comments or documentation content is described before the data is processed by sub-agents.
Audit Metadata