The Agent Skills Directory

[SAFE]: The skill instructions and scripts do not contain any malicious patterns, obfuscation, or unauthorized access attempts. All functionalities align with the stated purpose of performing biological pathway analysis using established bioinformatics libraries like gseapy and gprofiler-official. Downloads of gene set libraries are performed from trusted academic domains including Enrichr, g:Profiler, and MSigDB.
[DATA_EXFILTRATION]: No unauthorized data transmission was detected. Network access is restricted to the retrieval of biological metadata from reputable scientific APIs. The skill does not access sensitive local files, credentials, or environment variables.
[REMOTE_CODE_EXECUTION]: The helper script (scripts/run_enrichment.py) uses standard Python libraries for data processing and visualization. It does not employ unsafe shell execution, remote script piping, or dynamic execution of untrusted code.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied datasets from local files, which constitutes a managed ingestion surface. 1. Ingestion points: The skill reads gene symbols and rankings from text and CSV files provided via command-line arguments in scripts/run_enrichment.py. 2. Boundary markers: Data ingestion is handled programmatically via pandas, avoiding direct interpolation of raw file content into agent instructions. 3. Capability inventory: Permissions are limited to file reading, statistical calculation, and specific network queries for biological metadata. 4. Sanitization: Inputs are cleaned for biological symbol validity (casing and whitespace), which naturally restricts the potential for malicious text to influence the analysis flow.

pathway-enrichment