The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructions direct the user to download a helper script (call.mjs) from the vendor's infrastructure (fittings.sh) and execute it locally using Node.js.
[CREDENTIALS_UNSAFE]: The documentation guides the user to provide a private key through the FITTINGS_PRIVATE_KEY environment variable for payment signing. While the skill advises using a throwaway wallet, this practice exposes the key to the local environment context.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes medical device enforcement reports from the openFDA API, which is an external and untrusted data source.
Ingestion points: Data returned from the https://fittings.sh/v1/devicerecall/search endpoint (originating from openFDA).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the prompt interpolation.
Capability inventory: The agent can perform network requests and execute the provided client script.
Sanitization: The skill does not implement specific sanitization or filtering for the external API response content.
[DYNAMIC_EXECUTION]: The call.mjs script performs dynamic imports of its payment and cryptographic dependencies (@x402/fetch, @x402/evm, viem) and includes a self-integrity check that fetches a remote hash from the vendor's server.

devicerecall-search