The Agent Skills Directory

[PROMPT_INJECTION]: The skill integrates tools (youSearch, youContents) that fetch external web content, creating an indirect prompt injection surface.
Ingestion points: The youSearch and youContents tools fetch data from external URLs as seen in SKILL.md and assets/reference.ts.
Boundary markers: The skill explicitly implements a systemPrompt in assets/reference.ts and mandates it in SKILL.md to establish a trust boundary.
Capability inventory: The agent utilizes youSearch and youContents tools within a LangChain environment.
Sanitization: The skill relies on model-based instructions (system prompt) to ignore instructions in tool data; no programmatic sanitization of HTML/Markdown is specified.
[EXTERNAL_DOWNLOADS]: The skill installs dependencies from the vendor and well-known providers.
Evidence: SKILL.md recommends installing @youdotcom-oss/langchain, @langchain/core, and langchain. These originate from the author's own scope and the official LangChain organization.
[COMMAND_EXECUTION]: Shell commands are provided to facilitate environment setup.
Evidence: SKILL.md provides npm install, bun add, yarn add, and pnpm add commands for dependency management.

ydc-langchain-integration