The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill performs network requests to established scientific and identifier services, including the EMBL-EBI Ontology Lookup Service (OLS4), Bioregistry, Identifiers.org, and ZOOMA. These calls are used to retrieve metadata and resolve compact identifiers for biological research data.
[DATA_EXFILTRATION]: Functional scripts read scientific labels from local files or standard input using the --input flag and transmit these strings to external APIs as query parameters. This behavior is the intended mechanism for mapping local research data to global controlled vocabularies.
[COMMAND_EXECUTION]: The skill operates by executing provided Python scripts (e.g., resolve_terms.py, validate_terms.py) to process data. These scripts are self-contained and rely exclusively on the Python standard library.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text data for term resolution. It mitigates potential confusion through a structured evidence chain:
Ingestion points: Data is received via command-line arguments or local files processed by scripts/_inputs.py.
Boundary markers: The system implements a resolution strategy ladder (exact, token, fulltext) that explicitly labels matches (e.g., exact_label, partial), preventing the agent from misinterpreting arbitrary text as a confirmed term.
Capability inventory: The skill uses urllib.request for API queries and Path.read_text for local file access.
Sanitization: Inputs are sanitized using normalize_label (folding case/whitespace) and urllib.parse.urlencode before being interpolated into URL paths and query strings.

ontology-term-resolution