scientific-brainstorming

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • PROMPT_INJECTION (SAFE): The skill uses instructional language for role-playing a research partner without any bypass markers or attempts to override safety filters.
  • DATA_EXFILTRATION (SAFE): No access to sensitive files or unauthorized network calls were detected.
  • EXTERNAL_DOWNLOADS (SAFE): No external packages or remote scripts are requested, downloaded, or executed.
  • COMMAND_EXECUTION (SAFE): The skill contains no shell commands, system-level operations, or subprocess calls.
  • INDIRECT_PROMPT_INJECTION (SAFE): While the skill references a local file ('references/brainstorming_methods.md'), it lacks any executable capabilities that could be exploited via malicious content in processed data.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:44 PM