The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill requires the user to provide a private key via the FITTINGS_PRIVATE_KEY environment variable to sign payment authorizations for the x402 protocol. Storing sensitive private keys in the agent's operating environment carries a risk of credential exposure.
[EXTERNAL_DOWNLOADS]: The skill instructions suggest downloading the client script (call.mjs) from the developer's domain (fittings.sh) if it is not present. Additionally, the script requires the installation of several external Node.js libraries, including viem, @x402/fetch, and @x402/evm.
[DYNAMIC_EXECUTION]: The call.mjs script performs a self-integrity check by fetching a hash from the developer's server (fittings.sh/skill/call.mjs.sha256) and comparing it to its own local hash. It also utilizes dynamic imports to load its core libraries at runtime.
[INDIRECT_PROMPT_INJECTION]: The skill processes data returned from an external API, which introduces a surface for indirect prompt injection.
Ingestion points: Data enters the agent context from the output of call.mjs, which fetches tax rates from fittings.sh.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the API response as untrusted data.
Capability inventory: The agent has the capability to execute shell commands (node) and perform network operations.
Sanitization: The client script performs basic type coercion for input arguments but does not sanitize or filter the content returned by the API before presenting it to the agent.

tax-us-state-sales-rate