The Agent Skills Directory

[DATA_EXPOSURE]: The skill interacts with the official openFDA API ("api.fda.gov") to fetch public regulatory records. It manages API authentication securely through environment variables and implements a local file-based cache ("fda_cache") to store responses, which is a standard performance optimization.- [REMOTE_CODE_EXECUTION]: No remote code execution patterns were detected. The skill relies on the standard "requests" library for all network operations and does not attempt to download or run external executable content.- [PROMPT_INJECTION]: The skill's instructions and reference documentation are strictly focused on FDA data analysis. There are no attempts to manipulate the agent's system prompt, bypass safety filters, or use adversarial role-play techniques.- [COMMAND_EXECUTION]: The code is restricted to HTTP requests and data manipulation. There are no instances of shell command execution, subprocess management, or unsafe usage of dynamic execution functions like "eval()" or "exec()".- [EXTERNAL_DOWNLOADS]: All network traffic is directed toward well-known government API endpoints ("api.fda.gov"). The skill does not perform any unauthorized downloads from untrusted sources.

alterlab-fda