scientific-critical-thinking
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The SKILL.md file contains instructions that attempt to override the agent's default behavior by mandating the use of a specific persona ('Nano Banana Pro') and requiring that scientific schematics be generated 'by default' for new documents.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a specific bash command (
python scripts/generate_schematic.py) with natural language arguments. This is presented as a default action for document creation. - [DYNAMIC_EXECUTION]: The instructions rely on a script located at
scripts/generate_schematic.pywhich is not included in the provided file list. This makes the execution behavior of the 'Visual Enhancement' feature unverifiable and potentially dangerous if the script performs unauthorized actions. - [INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to ingest and evaluate untrusted external data, such as research papers and scientific claims. Ingestion points: Research papers, study protocols, and scientific claims (defined in SKILL.md). Boundary markers: The instructions lack explicit boundary markers to separate untrusted data from the agent's instructions. Capability inventory: The skill has access to Read, Write, Edit, and Bash tools, and is explicitly instructed to run a Python script. Sanitization: There are no instructions for sanitizing or escaping the content of the analyzed documents before processing.
Audit Metadata