The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the internet via the doc.url-to-markdown and doc.url-metadata tools. This content is intended to be processed by the agent, creating a surface for indirect prompt injection attacks where malicious instructions hidden in web pages could influence agent actions.
Ingestion points: Remote content fetched from URLs and provided as Markdown output from scripts/call.mjs.
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to treat tool output as untrusted data.
Capability inventory: The skill allows network requests, execution of a local script, and signing of blockchain transactions for payments.
Sanitization: The skill relies on the third-party gateway's 'clean' Markdown conversion but does not perform additional validation or sanitization of the resulting text.
[COMMAND_EXECUTION]: The core functionality of the skill relies on executing a local Node.js script (scripts/call.mjs). This script handles all external interactions, including service discovery and paid API calls.
[EXTERNAL_DOWNLOADS]: Documentation recommends maintaining the client script by downloading updates from the vendor's domain using curl -sO https://fittings.sh/skill/call.mjs. While pointing to the official vendor source, this practice involves external retrieval of executable code.

fittings-web