The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill performs network requests to well-known scientific and academic service providers.\n
Targeted domains include ebi.ac.uk (European Bioinformatics Institute), bioregistry.io, identifiers.org, and ontobee.org.\n
These requests are used to query biological ontology data and are handled via the standard library urllib.request module.\n
The domains are established research infrastructures and are treated as trusted sources for scientific metadata.\n- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of external data, creating a surface for indirect prompt injection.\n
Ingestion points: resolve_terms.py, map_terms.py, and validate_terms.py read user-provided labels and CURIEs from CLI arguments or files, and ingest JSON data from the OLS, Bioregistry, and ZOOMA APIs.\n
Boundary markers: The instructions in SKILL.md and references/curation-rules.md explicitly direct the agent to validate all candidates and check primary labels/definitions before accepting any term, mitigating the risk of following malicious instructions embedded in data.\n
Capability inventory: The skill is capable of network GET requests and writing data to files via standard Python file I/O. It does not contain capabilities for dynamic code execution (eval, exec) or system command invocation.\n
Sanitization: Input labels are normalized (case-folding and whitespace handling), and all dynamic API query parameters are URL-encoded. Output is consistently formatted as TSV or JSON.

ontology-term-resolution