egm-review

Pass

Audited by Gen Agent Trust Hub on Nov 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • File Access and Path Resolution: The skill identifies and resolves links within target documents. It includes a security gate that requires explicit user approval before accessing files located outside the current working directory, which helps prevent unauthorized file access.
  • External Content Downloads: To support review context, the skill can download content from external URLs. This process includes a requirement for user confirmation and saves the content locally, which is a standard part of the document aggregation workflow.
  • Indirect Prompt Injection Surface: The skill processes external, potentially untrusted document content. While this creates a surface for indirect instructions, the skill uses strict execution constraints—such as mandatory file-based delivery of reports and specific reviewer personas—to maintain control over the agent's behavior.
  • Subagent Delegation: The workflow relies on spawning independent subagents for different review phases. This protocol ensures that each review phase starts with a fresh context, reducing the risk of context contamination across review stages.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 3, 2026, 09:24 AM