The Agent Skills Directory

Data Exposure & Exfiltration (SAFE): All network communication is restricted to legitimate NCBI API endpoints (ncbi.nlm.nih.gov). The scripts do not access sensitive local files or hardcode API keys, opting for user-provided command-line arguments instead.
Unverifiable Dependencies & Remote Code Execution (SAFE): The skill relies exclusively on Python standard libraries (urllib, json, argparse). No external packages are installed, and no remote script execution patterns (e.g., curl piped to bash) were found.
Indirect Prompt Injection (SAFE): The skill processes external biological data from NCBI APIs and user-provided gene list files, creating an ingestion surface that is handled safely.
Ingestion points: scripts/batch_gene_lookup.py (file reading), and CLI parameters for gene symbols and IDs across all scripts.
Boundary markers: Absent; there are no explicit instructions for the agent to ignore embedded commands in the data.
Capability inventory: Network requests to NCBI via urllib.request and local file output via batch_gene_lookup.py.
Sanitization: All query inputs are URL-encoded using urllib.parse.urlencode to prevent malformed requests.

gene-database