The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including gene lists and expression datasets (TSV files), which are then analyzed via the Reactome API. The resulting JSON output is processed by the agent, creating a potential surface for indirect prompt injection if the source data or API response contains malicious instructions.
Ingestion points: The scripts/reactome_query.py script reads user-provided files (e.g., gene_list.txt) using open().read(). The agent also processes JSON responses from reactome.org.
Boundary markers: No specific delimiters or "ignore instructions" warnings are present in the scripts when processing this data.
Capability inventory: The skill possesses file read/write capabilities (open) and network operation capabilities (requests).
Sanitization: The reactome_query.py script applies basic string stripping (line.strip()) to input identifiers, but does not perform comprehensive sanitization against prompt injection patterns.
[EXTERNAL_DOWNLOADS]: The skill documentation recommends the installation of the reactome2py Python package via uv pip install reactome2py. This is the official client library for the Reactome database and is considered a standard dependency for this domain.

reactome-database