The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill interacts with the PubChem API at pubchem.ncbi.nlm.nih.gov, which is a well-known and trusted repository of chemical information maintained by the National Institutes of Health (NIH). All network operations are directed to this authoritative service.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes structured chemical data from an external source. While this creates a potential surface for indirect prompt injection if the database content were manipulated, the risk is minimal given the trusted nature of the source and the use of JSON parsing.\n
Ingestion points: Data retrieved via scripts/pubchem_api.py and stored in JSON output files in the workspace.\n
Boundary markers: Data is processed as structured JSON, though no specific instructions are provided to the LLM to ignore embedded commands within the chemical data.\n
Capability inventory: The skill uses uv run to execute local scripts and performs file writes to record license notifications and API results.\n
Sanitization: The wrapper script performs basic validation of HTTP response types and handles errors (400, 404) to prevent processing of malformed data.\n- [COMMAND_EXECUTION]: The skill utilizes uv run to execute a local helper script (scripts/pubchem_api.py). This is a legitimate use of command execution to provide a controlled interface for API interaction, ensuring rate limit compliance and robust error handling.

pubchem-database