The Agent Skills Directory

[DATA_EXFILTRATION] (SAFE): The scripts molecular_properties.py and similarity_search.py interact only with local file systems for reading molecule data and writing CSV results. No network libraries are imported, and no attempts to access sensitive system files (e.g., ~/.ssh, ~/.aws) were detected.
[REMOTE_CODE_EXECUTION] (SAFE): The skill does not download or execute remote scripts. It uses static Python code and established chemical informatics libraries. No dynamic code execution functions like eval() or exec() are used.
[COMMAND_EXECUTION] (SAFE): No subprocess calls or shell commands are performed. The skill uses standard file I/O and CLI argument parsing via argparse.
[PROMPT_INJECTION] (SAFE): Analysis of the markdown and Python files shows no instructions designed to override agent behavior, bypass safety filters, or extract system prompts.
[OBFUSCATION] (SAFE): All code and documentation are provided in plain text with no evidence of encoding (Base64), hidden characters (Zero-width), or homoglyph attacks.
[INDIRECT_PROMPT_INJECTION] (SAFE): While the skill ingests external SMILES and SDF files, this data is strictly processed as chemical structures by the RDKit library. There is no logic that interpolates this untrusted data into instructions or prompts for the LLM.

rdkit