Docs for LLMs

Feed an LLM or coding agent the yield.xyz docs through llms.txt, skill.md, or live MCP tools.

This documentation is published in three machine-readable forms an agent can consume without a human copying and pasting: a static index, a static full-text dump, and a set of live tools on the MCP server. Pick the one that fits how your agent works.

llms.txt and llms-full.txt

Every yield.xyz docs page is also available as plain text, following the llms.txt standard:

llms.txt

A directory of every page: title, one-line description, and URL. Lightweight; good for link discovery or building a retrieval index.

llms-full.txt

The full content of every page concatenated into one file. Upload it when you want an LLM to have the whole documentation as context in a single pass.

Both are generated from the live site, so they reflect whatever is currently published; there’s nothing to regenerate on your side.

skill.md and agent skills

https://docs.yield.xyz/skill.md is the root capability file for an agent that supports the Agent Skills standard. It describes the yield.xyz API, its authentication model, and points to the individual skills under skills/; each a self-contained instruction set for one job (find-yields, get-yield-info, enter-position, exit-position, manage-position, check-portfolio, programmatic-access). For the AgentKit skills that ship with the Claude Code plugins, see Agent skills.

MCP documentation tools

A static file is a snapshot; the MCP server gives an agent the same information live, mid-conversation, without pre-loading anything. Five tools are dedicated to this:

Tool	Use it to
yield_lookup_docs	Search the live docs index for pages relevant to a question
yield_fetch_doc	Read the full markdown of a specific docs.yield.xyz page
yield_get_api_spec	Query the live OpenAPI spec for exact field names, constraints, and error shapes
yield_troubleshoot_error	Diagnose an HTTP status or error message against the live spec
yield_list_repos	Find working reference code across yield.xyz’s public repos

These are the right choice for an agent that already has MCP access and needs an answer grounded in the current spec, rather than a document dumped into its context up front. Full inputs and outputs are in the tool reference.

Per-page contextual menu

Every page on this site carries a contextual menu (top right) for pulling that one page into another tool: copy the page as markdown, view it as raw markdown, open it in ChatGPT, Claude, or Perplexity with the page pre-loaded, or add these docs to Cursor, VS Code, or an MCP-compatible client. Reach for it when you want one page rather than the whole site; llms-full.txt and the MCP tools above are the better choice for broader context.

Next steps

Agent skills

Install the AgentKit skills into Claude Code, Codex, or Gemini CLI.

MCP server

Connect an agent to live yield data and transaction building.

Guardrails

Rules an agent follows before it signs or broadcasts anything.

Yield API reference

The REST API these docs and tools describe.