Sumi Docs MCP

Sumi Docs MCP

Read-only MCP server that exposes the curated Sumi Docs corpus through four tools, enabling agents to query documentation content programmatically.

Category
Visit Server

README

Sumi Docs

English | 简体中文

Sumi Docs publishes one reviewed documentation corpus for two consumers:

  • people browse an Astro and Starlight website;
  • agents query the same corpus through a read-only MCP server.

The reviewed root docs/ tree and content catalog are the semantic source of truth. The Web site and MCP server are independently addressable projections; an agent host is the MCP client.

The source repository and documentation site are public and under active development. No npm package, tagged GitHub Release, or supported binary has been published.

Prerequisites

  • Node.js 25.5.0 or newer
  • pnpm 10.26.0 through Corepack or the version declared in packageManager
  • a trusted checkout when loading repository-provided agent host configuration

First Run

pnpm install --frozen-lockfile
pnpm run build
node packages/mcp/dist/index.js doctor --json

Doctor reports project-relative paths or external-source placeholders by default. Add --show-paths only for local diagnosis; do not attach that output to public issues or build artifacts.

The checked-in sumi-docs.config.json selects root docs/ and the example OpenAPI document. Start the human site with:

pnpm --filter @sumi-os/docs-web dev

Open http://127.0.0.1:4321. Codex, Claude Code, and VS Code project adapters are described in Agent host integration. They expose the four MCP tools without requiring the optional maintainer Skill.

To expose the same corpus on a loopback Streamable HTTP endpoint:

node packages/mcp/dist/index.js serve --transport streamable-http

Connect a compatible MCP client to http://127.0.0.1:3000/mcp. The static Web site and _mcp corpus projection do not themselves run that endpoint.

For a hardened local container using the same reviewed root corpus:

docker compose up --build

The service exposes MCP at http://localhost:3000/mcp, liveness at /healthz, and corpus readiness at /readyz. Docker is optional for source development; public deployment still requires HTTPS termination and explicit Host policy.

The repository also provides three optional project workflows under .agents/skills/: $sumi-docs-use for setup and operation, $sumi-docs-pr for proposal and pull-request preparation, and $sumi-docs-audit for read-only, evidence-bound repository and release audits. They route work but do not replace MCP or publish remote changes.

Workspace

apps/web/                   Astro/Starlight site and corpus publisher
packages/mcp/               stdio and Streamable HTTP MCP server and CLI
packages/corpus-contract/   manifest schemas and conformance fixtures
docs/                       product handbook and default corpus
.agents/skills/             optional project usage and contribution workflows
Mode Command Purpose
Web development pnpm --filter @sumi-os/docs-web dev Local browser site with reload
MCP development pnpm --filter @sumi-os/docs-mcp dev TypeScript server against its example corpus
Production build pnpm run build Build contract, MCP, and static site
Compiled MCP node packages/mcp/dist/index.js serve Serve the discovered project corpus over stdio
Remote MCP endpoint node packages/mcp/dist/index.js serve --transport streamable-http Serve the same corpus on loopback HTTP
Validation pnpm run verify Package quality, tests, and dependency gates
Cross-product validation pnpm run verify:integration Exercise the generated Web corpus through MCP

Local operation and the default Pages-only publication require no runtime secrets. SITE_URL is required only for a release-site candidate. Enabling the optional production Streamable HTTP service requires reviewed endpoint variables and SSH material in the protected production-mcp environment. The container accepts documented SUMI_DOCS_* deployment variables; generated output, local state, logs, caches, and .env files remain ignored.

Package-specific instructions remain in each workspace. Active architecture decisions live in the owning workspace's docs/decisions/ directory; the root handbook presents their user-facing consequences without duplicating the decision records.

Source and site visibility do not publish a tag, package, or binary. Those artifacts remain subject to the documented release gates.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured