requesting-code-review
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent and its subagents to execute standard
gitcommands, such asgit rev-parse,git log,git diff, andgit worktree, to analyze code changes and manage repository history during the review process. - [PROMPT_INJECTION]: The subagent template in
code-reviewer.mdis susceptible to indirect prompt injection. It incorporates potentially untrusted data into its instruction prompt. - Ingestion points: Data is ingested via the
[DESCRIPTION]and[PLAN_OR_REQUIREMENTS]placeholders within thecode-reviewer.mdtemplate. - Boundary markers: The skill uses Markdown headers to partition the prompt, but it lacks robust delimiters or explicit instructions for the subagent to disregard commands that might be embedded in the implementation description or requirements.
- Capability inventory: The subagent is granted the ability to run shell commands for repository inspection and can create new directory structures using
git worktree. - Sanitization: There is no evidence of sanitization, escaping, or validation of the input text before it is interpolated into the instructions for the subagent.
Audit Metadata