scientific-problem-selection

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFENO_CODEEXTERNAL_DOWNLOADS
Full Analysis
  • [SAFE]: The skill implements a robust conversational framework for scientific ideation and project planning. No malicious patterns, prompt injections, or unauthorized access attempts were identified during the security audit. The instructions are aligned with the stated purpose of assisting researchers with strategic decisions.
  • [NO_CODE]: This skill package consists solely of Markdown instruction files and reference documentation. There are no Python or Node.js scripts, shell commands, or binary files included in the package, which eliminates the risk of unauthorized local code execution.
  • [EXTERNAL_DOWNLOADS]: The skill documentation describes the use of PubMed searches for literature verification and benchmarking. PubMed is a well-known and authoritative scientific database service. The instructions specify information retrieval from this source to support research validation, which is a safe and standard functional requirement for a scientific assistant.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 11:31 PM
Security Audit — agent-trust-hub — scientific-problem-selection