brainstorming-research

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided data, which constitutes an indirect prompt injection surface.
  • Ingestion points: In SKILL.md, the agent prompts the user to provide existing materials such as paper titles, abstracts, or supervisor-mandated structures during the "已有论文信息提示" (Existing Paper Information Prompt) phase.
  • Boundary markers: The skill does not implement specific delimiters or "ignore embedded instructions" warnings when interpolating this user data into generated project files.
  • Capability inventory: The skill's primary capabilities involve file system operations, specifically creating the plan/ and chapters/ directories and writing markdown files (project-overview.md, outline.md, progress.md) in SKILL.md under the "创建项目结构" (Create Project Structure) section.
  • Sanitization: No explicit sanitization or validation logic is defined for the user-supplied research content before it is summarized and written to disk.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 01:57 AM
Security Audit — agent-trust-hub — brainstorming-research