The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses sensitive local file paths to retrieve blockchain private keys, specifically searching ~/.mcporter/mcporter.json and ~/.x402-config.json. While this is required for the skill's stated purpose of automated payments, accessing credentials from other applications' configuration files is a significant security risk.
[COMMAND_EXECUTION]: The skill executes a bundled JavaScript tool (dist/x402_invoke.js) that performs blockchain signing and transaction submission. This tool has access to the filesystem and network to carry out its operations.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it fetches and returns responses from external, user-supplied URLs to the agent without boundary markers.
Ingestion points: Data enters the agent context via the X402FetchClient.request method which retrieves JSON, text, or binary data from the target endpoint (src/x402_invoke.ts).
Boundary markers: There are no explicit delimiters or system instructions used to separate the external response content from the agent's core logic.
Capability inventory: The skill can sign and broadcast blockchain transactions, write temporary files to /tmp, and read sensitive configuration files.
Sanitization: The skill implements a key redaction mechanism that filters private keys from error messages and stack traces using regular expressions before they are output to the console.

x402-payment