io.github.andrea9293/mcp-documentation-server MCP Server
io.github.andrea9293/mcp-documentation-server
Local-first document management and semantic search for AI coding agents—no cloud, no vendor lock-in.
What this kind of tool is for
- Automate a build, test, or deployment step you repeat by hand.
- Review or refactor code in an existing repository without leaving your agent.
- Bridge an internal service or CLI into the agent toolset.
How to compare servers like this
- Scope of control:
- Repos and CI are hard to undo; prefer tools that ask before anything irreversible.
- Environment fit:
- Match the runtime it expects (Node, Python, a container) to what you actually run.
- Maintenance signal:
- Check the last-commit date — dev tooling that lags upstream breaks first.
- Config surface:
- Prefer tools whose environment-variable or config keys are documented plainly.
This page carries your copy-paste config: pick your MCP client in the install section on this page and use the copy button.
What is the io.github.andrea9293/mcp-documentation-server MCP server?
The MCP Documentation Server is a local-first document management system that provides semantic search capabilities for AI coding agents using Orama vector database and local embeddings. It runs fully offline with a built-in web dashboard and REST API, supporting hybrid full-text and vector search across documents.
This server lets you build and query a knowledge base entirely on your machine. Upload documents (txt, md, pdf), search semantically across them using local embeddings, and expose the results to AI agents via MCP tools or REST API. It includes a web UI for browsing and managing documents, optional AI-powered search via Google Gemini, and parent-child chunking for better context retrieval.
How to install io.github.andrea9293/mcp-documentation-server
Copy-paste configuration for popular MCP clients.
{
"mcpServers": {
"mcp-documentation-server": {
"command": "npx",
"args": [
"-y",
"@andrea9293/mcp-documentation-server"
]
}
}
}Tools & capabilities
Tools this server exposes to the agent.
add_document— Add a document with title, content, and optional metadatalist_documents— List all documents with metadata and content previewget_document— Retrieve the full content of a document by IDdelete_document— Remove a document, its chunks, database entries, and associated filesprocess_uploads— Process all files in the uploads folder (chunking + embeddings)get_uploads_path— Returns the absolute path to the uploads folderlist_uploads_files— Lists files in the uploads folder with size and format infoget_ui_url— Returns the Web UI URLsearch_documents— Semantic vector search within a specific documentsearch_all_documents— Hybrid (full-text + vector) cross-document searchget_context_window— Returns a window of chunks around a given chunk indexsearch_documents_with_ai— AI-powered search using Google Gemini (requires GEMINI_API_KEY)
Use cases
- Build a local knowledge base from documentation, code snippets, or research papers and query it semantically without external APIs
- Integrate semantic search into AI coding agents to provide context-aware answers from your own documents
- Upload PDF, Markdown, and text files via drag-and-drop and search across them with hybrid full-text and vector similarity
- Use the REST API to expose document search to multiple AI agents (Claude, Cursor, Gemini CLI) with a single server instance
- Analyze documents with optional AI-powered search using Google Gemini for advanced reasoning over your knowledge base
io.github.andrea9293/mcp-documentation-server MCP server FAQ
It's a local-first document management and semantic search system for AI agents. It runs entirely offline using Orama vector database and local embeddings (Transformers.js), with an optional web UI and REST API.
Yes, it's open-source under the MIT license. The embedding models are free. Optional AI search via Google Gemini requires your own API key.
Add it to your MCP configuration as: {"command": "npx", "args": ["-y", "@andrea9293/mcp-documentation-server"]}. The web UI starts automatically on http://localhost:3080.
No. Everything runs locally on your machine. The only optional external dependency is Google Gemini API for AI-powered search, which you can skip.
The server supports .txt, .md, and .pdf files. You can upload them via the web UI drag-and-drop or place them in the uploads folder.
It uses hybrid search combining full-text indexing and vector similarity. Documents are split using parent-child chunking to preserve context, and embeddings are generated locally using Transformers.js.
README (reference)
Source of truth, from the repository.
MCP Documentation Server
Local-first document management and semantic search for AI coding agents. No external databases, no cloud APIs, no vendor lock-in.
Unlike other MCP servers that are CLI-only, this one ships with a full web dashboard — browse, search, upload, and manage your knowledge base from your browser. Every MCP tool is also exposed as a REST API, giving AI agents a lean, schema-free interface.
- 🏠 Runs fully offline — Orama vector DB with local AI embeddings (Transformers.js)
- 🌐 Built-in Web UI — starts automatically on port 3080 alongside the MCP server
- 🔍 Hybrid search — full-text + vector similarity with parent-child chunking
- 🤖 Optional AI search — Google Gemini for advanced document analysis (bring your own key)
- 📁 Drag & drop uploads —
.txt,.md,.pdfsupport - 📦 Published on the MCP Registry — installable via npx, no clone needed
Quick Start
{
"mcpServers": {
"documentation": {
"command": "npx",
"args": ["-y", "@andrea9293/mcp-documentation-server"]
}
}
}
Open your browser at http://localhost:3080 — the web UI starts automatically.
🤖 Agent Skill (REST API) — recommended for AI agents
Every MCP tool is also accessible via the REST API on http://127.0.0.1:3080/api/. This is the recommended way to interact from AI agents (Claude Code, OpenCode, Gemini CLI, Cursor) because it avoids loading MCP tool schemas into the conversation context — only the response JSON enters.
curl -s http://127.0.0.1:3080/api/config
curl -s http://127.0.0.1:3080/api/documents
curl -s -X POST http://127.0.0.1:3080/api/search-all \
-H "Content-Type: application/json" \
-d '{"query": "your search", "limit": 5}'
A ready-to-use skill is included at skills/documentation-server/SKILL.md — it teaches your agent every endpoint with examples. Install it:
npx skills add https://github.com/andrea9293/mcp-documentation-server --skill documentation-server
Basic workflow
- Add documents using
add_documentor place.txt/.md/.pdffiles in the uploads folder and callprocess_uploads. - Search across everything with
search_all_documents, or within a single document withsearch_documents. - Use
get_context_windowto fetch neighboring chunks and give the LLM broader context.
Web UI
The web interface starts automatically on port 3080 when the MCP server launches. From the web UI you can:
- 📊 Dashboard — overview of all documents and stats
- 📄 Documents — browse, view, and delete documents
- ➕ Add Document — create documents with title, content, and metadata
- 🔍 Search All — semantic search across all documents
- 🎯 Search in Doc — search within a specific document
- 🤖 AI Search — Gemini-powered analysis (if
GEMINI_API_KEYis set) - 📁 Upload Files — drag & drop files and process them into the knowledge base
- 🪟 Context Window — explore chunks around a specific index
Configure an MCP client
Minimal
{
"mcpServers": {
"documentation": {
"command": "npx",
"args": ["-y", "@andrea9293/mcp-documentation-server"]
}
}
}
With environment variables (all optional)
{
"mcpServers": {
"documentation": {
"command": "npx",
"args": ["-y", "@andrea9293/mcp-documentation-server"],
"env": {
"MCP_BASE_DIR": "/path/to/workspace",
"GEMINI_API_KEY": "your-api-key-here",
"MCP_EMBEDDING_MODEL": "Xenova/all-MiniLM-L6-v2",
"START_WEB_UI": "true",
"WEB_HOST": "127.0.0.1",
"WEB_PORT": "3080"
}
}
}
}
All environment variables are optional. Without GEMINI_API_KEY, only the local embedding-based search tools are available.
MCP Tools
The server registers the following tools (all validated with Zod schemas):
📄 Document Management
| Tool | Description |
|---|---|
add_document | Add a document (title, content, optional metadata) |
list_documents | List all documents with metadata and content preview |
get_document | Retrieve the full content of a document by ID |
delete_document | Remove a document, its chunks, database entries, and associated files |
📁 File Processing
| Tool | Description |
|---|---|
process_uploads | Process all files in the uploads folder (chunking + embeddings) |
get_uploads_path | Returns the absolute path to the uploads folder |
list_uploads_files | Lists files in the uploads folder with size and format info |
get_ui_url | Returns the Web UI URL (e.g. http://localhost:3080) — useful to open the dashboard or to locate the uploads folder from the browser |
🔍 Search
| Tool | Description |
|---|---|
search_documents | Semantic vector search within a specific document |
search_all_documents | Hybrid (full-text + vector) cross-document search |
get_context_window | Returns a window of chunks around a given chunk index |
search_documents_with_ai | 🤖 AI-powered search using Gemini (requires GEMINI_API_KEY) |
Configuration
Configure via environment variables or a .env file in the project root:
| Variable | Default | Description |
|---|---|---|
MCP_BASE_DIR | ~/.mcp-documentation-server | Base directory for data storage |
MCP_EMBEDDING_MODEL | Xenova/all-MiniLM-L6-v2 | Embedding model name |
GEMINI_API_KEY | — | Google Gemini API key (enables search_documents_with_ai) |
MCP_CACHE_ENABLED | true | Enable/disable LRU embedding cache |
START_WEB_UI | true | Set to false to disable the built-in web interface |
WEB_HOST | 127.0.0.1 | Bind address for the web UI (use 0.0.0.0 to expose on all interfaces) |
WEB_PORT | 3080 | Port for the web UI |
MCP_STREAMING_ENABLED | true | Enable streaming reads for large files |
MCP_STREAM_CHUNK_SIZE | 65536 | Streaming buffer size in bytes (64KB) |
MCP_STREAM_FILE_SIZE_LIMIT | 10485760 | Threshold to switch to streaming (10MB) |
Storage layout
~/.mcp-documentation-server/ # Or custom path via MCP_BASE_DIR
├── data/
│ ├── orama-chunks.msp # Orama vector DB (child chunks + embeddings)
│ ├── orama-docs.msp # Orama document DB (full content + metadata)
│ ├── orama-parents.msp # Orama parent chunks DB (context sections)
│ ├── migration-complete.flag # Written after legacy JSON migration
│ └── *.md # Markdown copies of documents
└── uploads/ # Drop .txt, .md, .pdf files here
Embedding Models
Set via MCP_EMBEDDING_MODEL:
| Model | Dimensions | Notes |
|---|---|---|
Xenova/all-MiniLM-L6-v2 | 384 | Default — fast, good quality |
Xenova/paraphrase-multilingual-mpnet-base-v2 | 768 | Recommended — best quality, multilingual |
Models are downloaded on first use (~80–420 MB). The vector dimension is determined automatically from the provider.
⚠️ Important: Changing the embedding model requires re-adding all documents — embeddings from different models are incompatible. The Orama database is recreated automatically when the dimension changes.
Architecture
Server (FastMCP, stdio)
├─ Web UI (Express, port 3080)
│ └─ REST API → DocumentManager
└─ MCP Tools
└─ DocumentManager
├─ OramaStore — Orama vector DB (chunks DB + docs DB + parents DB), persistence, migration
├─ IntelligentChunker — Parent-child chunking (code, markdown, text, PDF)
├─ EmbeddingProvider — Local embeddings via @xenova/transformers
│ └─ EmbeddingCache — LRU in-memory cache
└─ GeminiSearchService — Optional AI search via Google Gemini
- OramaStore manages three Orama instances: one for document metadata/content, one for child chunks with vector embeddings, and one for parent chunks (context sections). All are persisted to binary files on disk and restored on startup.
- IntelligentChunker implements the Parent-Child Chunking pattern: documents are first split into large parent chunks that preserve full context (sections, paragraphs), then each parent is further split into small child chunks for precise vector search. At query time, results are deduplicated by parent so that the LLM receives both the matched fragment and the broader context.
- EmbeddingProvider lazily loads a Transformers.js model for local inference — no API calls needed.
Development
git clone https://github.com/andrea9293/mcp-documentation-server.git
cd mcp-documentation-server
npm install
npm run dev # FastMCP dev mode with hot reload
npm run build # TypeScript compilation
npm run inspect # FastMCP web UI for interactive tool testing
npm start # Direct tsx execution (MCP server + web UI)
npm run web # Run only the web UI (development)
npm run web:build # Run only the web UI (compiled)
Contributing
- Fork the repository
- Create a feature branch:
git checkout -b feature/name - Follow Conventional Commits for messages
- Open a pull request
License
MIT — see LICENSE
Support
- 📖 Documentation
- 🐛 Report Issues
- 💬 MCP Community
- 🤖 Google AI Studio — get a Gemini API key
Star History
Related MCP servers

Codex Delegate
Bridge AI coding hosts to the OpenAI Codex CLI for delegated implementation.

Cursor Delegate
Bridge AI coding hosts to cursor-agent over the Agent Client Protocol.

Binky
Isolated dev environments, one per git worktree: ports, env, and DB slices as agent tools.
Behavioral drift context (Behavioral Load Map, hot paths, release brief) for coding agents, per PR.

Anomalia
AI social autopilot: plan, write, design and publish posts, blog/SEO, news radar and leads.

MCP server for Google Flights: search, price tracking, cabin comparison, emissions, and more.

