The Agent Skills Directory

[SAFE]: External dependency. The skill utilizes the tooluniverse Python package which originates from a recognized scientific research institution (Zitnik Lab at Harvard). The installation process via uv pip install is a standard and expected procedure for this type of agent skill.
[SAFE]: Credential Handling. The documentation for the OpenRouterClient correctly instructs users to provide their own API keys and uses placeholders like your_key in code examples, avoiding the use of hardcoded secrets.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill provides tools to fetch and analyze data from external scientific sources (e.g., PubMed, UniProt, and chemical databases) which is then processed by LLM-based components.
Ingestion points: External data enters the context via tools like PubMed_search and GEO_download_dataset in scripts/example_workflow.py and references/tool-composition.md.
Boundary markers: The instructions do not specify the use of delimiters or specific system instructions to ignore potential commands embedded within the retrieved scientific text.
Capability inventory: The skill allows for network requests (scientific API access), file system interaction (via result-saving hooks), and complex multi-step workflow orchestration.
Sanitization: The skill documentation recommends validating scientific identifiers such as SMILES strings and UniProt IDs, which serves as a security control for those specific data types.

tooluniverse