scientific-critical-thinking

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to utilize the Bash tool to run a local Python script (scripts/generate_schematic.py) for the purpose of creating scientific diagrams and schematics.
  • [EXTERNAL_DOWNLOADS]: The skill references an external URL (www.k-dense.ai) which is presented as a hosted research platform associated with the skill's author, K-Dense Inc.
  • [PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it is designed to ingest and analyze external, potentially untrusted scientific documents and claims.
  • Ingestion points: Scientific research papers, study protocols, and external claims analyzed via the Read tool.
  • Boundary markers: The instructions do not specify the use of delimiters or 'ignore' warnings when processing external text content.
  • Capability inventory: The skill allows access to Bash (for local script execution) and file modification via Write and Edit tools.
  • Sanitization: No specific content sanitization or instruction-stripping logic is defined for the analyzed data.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 11:30 PM
Security Audit — agent-trust-hub — scientific-critical-thinking