sp-requesting-code-review

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructions and subagent templates include shell commands for interacting with the git version control system. Specifically, it uses git rev-parse, git log, git diff, and git worktree to analyze code changes and manage temporary review environments.
  • [INDIRECT_PROMPT_INJECTION]: The code-reviewer.md template defines a process where external content (implemented descriptions and plan requirements) is interpolated directly into a subagent's prompt. This creates a surface where malicious instructions embedded in a project's documentation or plan files could influence the subagent's behavior.
  • Ingestion points: The subagent prompt in code-reviewer.md ingests [DESCRIPTION] and [PLAN_OR_REQUIREMENTS] variables provided from the project context.
  • Boundary markers: No specific delimiters or "ignore instructions" warnings are present around the interpolated fields to prevent the subagent from following instructions found within those variables.
  • Capability inventory: The subagent is granted capabilities to execute git commands and load the scrutinize-dispatch skill.
  • Sanitization: No evidence of escaping or validation of the input variables is present before interpolation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 02:26 AM
Security Audit — agent-trust-hub — sp-requesting-code-review