requesting-code-review
Pass
Audited by Gen Agent Trust Hub on Apr 28, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Indirect prompt injection surface identified in the
code-reviewer.mdtemplate. - Ingestion points: The reviewer subagent ingests potentially untrusted data through the
{WHAT_WAS_IMPLEMENTED},{PLAN_OR_REQUIREMENTS}, and{DESCRIPTION}placeholders. - Boundary markers: There are no explicit delimiters or boundary markers (such as XML tags or unique string wrappers) to separate this external context from the subagent's instruction logic.
- Capability inventory: The subagent is granted the capability to execute shell commands, specifically
git diffoperations, as part of the review process. - Sanitization: The skill does not implement validation or sanitization for the input data provided to the subagent's template.
Audit Metadata