The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates data retrieval from a wide array of well-known and reputable scientific and governmental services, including NASA, NCBI, the World Bank, and SEC EDGAR. These interactions are standard for the skill's purpose of scientific and demographic data lookup.
[DYNAMIC_EXECUTION]: For the BRENDA database, which utilizes a SOAP API instead of REST, the skill instructs the agent to generate and execute a Python script using the zeep library. This is a legitimate requirement for interacting with that specific data source and is scoped to the skill's primary function.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest third-party data from various public APIs. To mitigate the risk of malicious instructions within that data, the instructions provide clear guidelines to the agent: treat all API responses as untrusted data, avoid following instructions embedded in payloads, and never pass raw response text into shell commands.
[SAFE_PRACTICE]: The skill manages API keys by instructing the agent to look up specific variables in the environment or a .env file narrowly. It explicitly forbids the agent from printing, logging, or revealing these secrets in its output or provenance records.

database-lookup