scientific-critical-thinking
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection. It is designed to ingest and interpret untrusted data from external sources such as research papers and media reports, which could contain hidden malicious instructions.
- Ingestion points: Untrusted data enters the agent context through the analysis of "research papers" and "media reports" as specified in "SKILL.md" under the "Claim Evaluation" section.
- Boundary markers: The instructions lack defined delimiters or specific warnings to the agent to disregard instructions embedded within the ingested data.
- Capability inventory: The agent has access to "Bash", "Write", "Edit", and "Read" tools, providing a significant capability surface if successfully exploited.
- Sanitization: No sanitization or validation of the input data is described in the prompt logic.
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute a local Python script via the shell to generate scientific schematics.
- Evidence: "python scripts/generate_schematic.py "your diagram description" -o figures/output.png" in "SKILL.md".
Audit Metadata