rdkit

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The documentation in SKILL.md provides a performance optimization example that uses pickle.load() to read molecular data from local files. Because the pickle module can execute arbitrary code during deserialization, this represents a security risk if the agent or user is convinced to load a maliciously crafted .pkl file.
  • [SAFE]: The skill implementation correctly utilizes the well-known and trusted RDKit library for legitimate cheminformatics tasks such as SMILES parsing, descriptor calculation, and substructure searching.
  • [SAFE]: The included Python scripts (molecular_properties.py, similarity_search.py, substructure_filter.py) follow best practices for CLI tool development, using local file I/O for processing and containing no logic for network communication, data exfiltration, or credential harvesting.
  • [SAFE]: No evidence of prompt injection, obfuscation, or unauthorized privilege escalation was detected in the skill's instructions or metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 11:38 AM
Security Audit — agent-trust-hub — rdkit