The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from local files, such as gene lists and expression TSV matrices, which are then processed and sent to the Reactome API. Ingestion points: The command_analyze function in scripts/reactome_query.py reads user-provided files, as do the code snippets in SKILL.md. Boundary markers: The skill does not define specific delimiters or instructions to prevent the agent from interpreting content within the ingested files as instructions. Capability inventory: The skill uses Bash(python:*) to run helper scripts, requests for network POST operations to reactome.org, and open() for local file reading and writing. Sanitization: The script performs basic whitespace stripping and regex-based HTML tag removal from API responses, but lacks validation for the ingested input data.\n- [COMMAND_EXECUTION]: The skill documentation instructs the agent to execute a Python helper script via the shell for various biological database operations. Evidence: SKILL.md provides multiple examples of running scripts/reactome_query.py with the uv tool to query version information, search for pathways, and analyze data.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation and use of a domain-specific Python package for pathway analysis. Evidence: SKILL.md includes installation instructions for the reactome2py library, which is the official Python client for the Reactome service.

alterlab-reactome