iterative-review

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes system commands to interact with the local Git repository. Specifically, it executes git diff and git log to extract context for the review and performs automated git commit operations to apply suggested fixes. This behavior is documented and central to the skill's purpose.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data (source code diffs) and provides it to an LLM subagent without robust sanitization.
  • Ingestion points: The output of git diff (or user-provided context) is interpolated directly into the senior reviewer's prompt in Step 1.
  • Boundary markers: The code diff is enclosed in markdown triple backticks, but the system framing lacks a specific directive for the subagent to ignore instructions or payload markers contained within the code itself.
  • Capability inventory: The agent can read repository history, write files, and execute commits to modify the codebase.
  • Sanitization: There is no evidence of sanitization, escaping, or filtering applied to the diff content before it is processed by the subagent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 11:53 PM