LLM Reference MCP server
Commit Oct 21, 2026Provides versioned documentation and semantic search for AI tools and coding assistants through an MCP endpoint.
Also known as reclear-io/llmref-plugin, llmref-plugin
- Stars
- 0
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- 0.1.0
- Last commit
- Oct 21, 2026
- License
- —
What is LLM Reference MCP?
LLM Reference provides versioned documentation and semantic search for AI tools and coding assistants through an MCP endpoint.
Summary generated from the project's README
About LLM Reference MCP server
Accessing versioned, model-ready documentation and semantic search through an AI agent is what this remote MCP server enables. Developers using Claude Code or other MCP-capable clients like Cursor and Claude Desktop can connect to the remote endpoint to retrieve precise technical information. The server provides specific tools including list_projects, list_versions, resolve_docs, and search_docs to help agents navigate documentation accurately.
This tool is designed to prevent agents from relying solely on outdated training data by providing a way to fetch version-specific details. It functions as a streamable-HTTP endpoint that connects to a central API. When permitted, the system can also use the report_missing_docs method to provide feedback on missing public documentation or official llms.txt files without ever sending private code, credentials, or conversation text.
What you can do with it
- 1Retrieve version-specific technical documentationAsk the agent to find the exact documentation for a specific version of a library instead of relying on its internal training data.
- 2Identify missing public documentationAllow the agent to report when it cannot find official documentation for a specific tool or project.
Written from the project's README. Every use case is checked against a quote from it.
How to install LLM Reference MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall LLM Reference MCP in Claude Code
$ claude mcp add --transport http llmref-plugin https://api.llmref.org/v1/mcpInstall LLM Reference MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"llmref-plugin": {
"url": "https://api.llmref.org/v1/mcp"
}
}
}Install LLM Reference MCP in VS Code
.vscode/mcp.json
{
"servers": {
"llmref-plugin": {
"type": "http",
"url": "https://api.llmref.org/v1/mcp"
}
}
}Install LLM Reference MCP in VS Code CLI
$ code --add-mcp '{"name":"llmref-plugin","type":"http","url":"https://api.llmref.org/v1/mcp"}'Install LLM Reference MCP in Codex
$ codex mcp add llmref-plugin --url https://api.llmref.org/v1/mcpInstall LLM Reference MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.llmref-plugin] url = "https://api.llmref.org/v1/mcp"
Install LLM Reference MCP in Gemini CLI
$ gemini mcp add -s user --transport http llmref-plugin https://api.llmref.org/v1/mcpInstall LLM Reference MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"llmref-plugin": {
"url": "https://api.llmref.org/v1/mcp",
"type": "http"
}
}
}Automated checks
Ran Nov 15, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachablereclear-io/llmref-plugin is public and reachablegithub.com/reclear-io/llmref-plugin ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)No licenseFinding
- Commit or release in the last yearLast commit or release 2026-10-21github.com/reclear-io/llmref-plugin ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeinitialize answered with protocol 2025-06-18 in 352 msPass
README
View on GitHubLLM Reference plugin
A Claude Code plugin that connects to LLM Reference - versioned, model-ready documentation and semantic search - and ships a skill that teaches an agent how and when to use it.
What you get
- MCP server
llmref(remote,https://api.llmref.org/v1/mcp) exposinglist_projects,list_versions,resolve_docs,search_docs, and the optionalreport_missing_docsfeedback method. - Skill
llmrefthat tells the agent to prefer LLM Reference over training data for version-specific docs, and the order to call the tools in.
Install (Claude Code)
/plugin marketplace add reclear-io/llmref-plugin
/plugin install llmref@llmref
Use in any other MCP client
The server is a plain remote streamable-HTTP MCP endpoint, so any MCP-capable client can connect directly:
https://api.llmref.org/v1/mcp
Per-client setup (Claude Desktop, Cursor, Codex, and others) is documented at https://llmref.org.
License
MIT - Open Source by Reclear.io
Optional discovery feedback
When permitted, the skill quietly reports missing public docs and any official llms.txt or docs URL already found during the task. It never waits for ingestion or sends conversation text, code, credentials or private URLs. Tell the agent not to report to opt out. Host permissions apply. See PRIVACY.md.
LLM Reference MCP questions
How do I add LLM Reference MCP to Claude Code?
Run claude mcp add --transport http llmref-plugin https://api.llmref.org/v1/mcp in your terminal. Add --scope user to make it available in every project.
How do I add LLM Reference MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
How can I use this with clients other than Claude Code?
The server is a remote streamable-HTTP MCP endpoint that any MCP-capable client like Claude Desktop, Cursor, or Codex can connect to directly.
What tools does the llmref MCP server expose?
The server exposes list_projects, list_versions, resolve_docs, search_docs, and an optional report_missing_docs method.
Does the server send my private code or credentials to the registry?
No, the feedback method never sends conversation text, code, credentials, or private URLs.