The Agent Skills Directory

[COMMAND_EXECUTION]: The skill operates by calling a local Node.js script (scripts/call.mjs) which handles command-line arguments to perform searches, describe services, and execute paid API calls.
[CREDENTIALS_UNSAFE]: The skill requires the FITTINGS_PRIVATE_KEY environment variable to be set with a blockchain private key. This key is used locally by the viem library to sign payment authorizations for the x402 protocol. The instructions provide explicit warnings against committing or printing the key, but its presence in the environment is a sensitive credential requirement.
[EXTERNAL_DOWNLOADS]: The skill documentation and scripts refer to external Node.js dependencies (viem, @x402/fetch, and @x402/evm) that must be installed by the user. It also includes a mechanism to verify the script's version against the vendor's server and suggests downloading updates via curl from https://fittings.sh/skill/call.mjs.
[PROMPT_INJECTION]: The skill processes medical records, trials, and nutrition data from external registries. While these are reputable US federal sources, the ingestion of this external content into the agent's context for processing (e.g., checking trial eligibility) creates a surface for indirect prompt injection.

health-reference