openzim-mcp
MCP server for reading and searching ZIM archives, optimized for single-user deployments with a persistent Wikipedia archive.
README
OpenZIM MCP Server
openzim-mcp is an MCP server for reading and searching ZIM archives.
This fork keeps the same package name but is intentionally opinionated for a single-user deployment:
- bare-metal, not Docker-first
systemd-managed HTTP service- simple mode as the default operating model
- one persistent Wikipedia archive at a known path
- reduced prompt surface:
/researchonly
Current fork version: 2.0.1
Maintainer: sibyl
What This Fork Changes
Compared with upstream, this fork is optimized for a narrow and stable setup.
- simple mode still exposes
zim_query - simple mode exposes only the
/researchMCP prompt summarizeandexploreprompts are not registered- when a client mistakenly passes an entry path into
zim_file_path, simple mode treats that as a slot error and falls back to the single loaded archive
That last point matters for small models and weak MCP clients. A payload like this:
{
"compact": true,
"query": "summary of A/Marcellina_(gnostic)",
"zim_file_path": "A/Marcellina_(gnostic)"
}
now resolves against the real archive instead of treating
A/Marcellina_(gnostic) as a filesystem path.
Intended Deployment
This fork is written around a single persistent archive:
/srv/zim/wikipedia_en_all_maxi_2022-05.zim
and a local HTTP service bound to loopback, typically behind a reverse proxy.
Example service command:
uv run openzim-mcp --mode simple --transport http --host 127.0.0.1 --port 3338 /srv/zim
Features Kept In This Fork
zim_querynatural-language interface- advanced mode codepath and tooling
- HTTP transport
- search, article retrieval, structure, links, title lookup
- compact simple-mode rendering
- single registered MCP prompt:
/research
Features Intentionally De-Emphasized
- Docker-first deployment guidance
- broad multi-client prompt workflows
- prompt flows that require the caller to supply
zim_file_path
The code still contains far more capability than this README emphasizes. The difference is intentional: this document is for operating this fork, not for advertising every upstream feature.
Installation
Requirements:
- Python
3.12+ uv- a readable directory containing your
.zimfiles
Clone the repo:
git clone <your-fork-url>
cd openzim-mcp
Install dependencies:
uv sync
Run locally:
uv run openzim-mcp --mode simple /srv/zim
Run over HTTP:
uv run openzim-mcp --mode simple --transport http --host 127.0.0.1 --port 3338 /srv/zim
Bare-Metal systemd Deployment
This fork is expected to run directly on a host, not in a container.
Example unit:
[Unit]
Description=OpenZIM MCP Server
After=network-online.target
Wants=network-online.target
[Service]
User=openzim-mcp
Group=openzim-mcp
WorkingDirectory=/opt/openzim-mcp/app
EnvironmentFile=/etc/openzim-mcp/openzim-mcp.env
ExecStart=/var/lib/openzim-mcp/.local/bin/uv run openzim-mcp --mode simple --transport http --host 127.0.0.1 --port 3338 /srv/zim
Restart=always
RestartSec=5
NoNewPrivileges=true
PrivateTmp=true
ProtectSystem=strict
ProtectHome=true
ReadOnlyPaths=/srv/zim /opt/openzim-mcp
ReadWritePaths=/var/lib/openzim-mcp
RestrictSUIDSGID=true
LockPersonality=true
[Install]
WantedBy=multi-user.target
Enable it:
sudo systemctl daemon-reload
sudo systemctl enable --now openzim-mcp
sudo systemctl status openzim-mcp
Updating The Deployment
For this setup, do not copy local virtual environments into the server tree.
Deploy source only:
- copy or pull the repo into
/opt/openzim-mcp/app - keep
.venvoff the server copy - rebuild the environment on the target host with
uv sync
Typical update flow:
sudo systemctl stop openzim-mcp
cd /opt/openzim-mcp/app
git pull
sudo -u openzim-mcp /var/lib/openzim-mcp/.local/bin/uv sync
sudo systemctl start openzim-mcp
If you are deploying by copying files rather than pulling on-host, replace the whole app tree rather than pasting selected files over an existing checkout.
MCP Surface In Simple Mode
Simple mode in this fork is intentionally small.
Tools:
zim_query
Prompts:
/research/article/compare/timeline/map
That is the interface you should expect clients to see.
All prompts are workflow macros over zim_query; they do not reference the
advanced-mode tools. They assume a single offline Wikipedia archive and tell
clients to omit zim_file_path so the server can auto-select the loaded
archive.
Example Queries
These are the kinds of simple-mode requests this fork is tuned for:
tell me about Marcellinasummary of A/Marcellina_(gnostic)show structure of Photosynthesislinks in Ada Lovelacefind article titled Hypatiasearch for gnosticism
Versioning
This fork now uses a normal release version:
2.0.1
That replaces the upstream alpha train version that was present in this tree.
Development Notes
Useful commands:
uv run pytest tests/test_server.py tests/test_simple_tools.py tests/test_prompts.py -q
uv run pytest -q
The fork-specific behavior currently pinned by tests includes:
- simple mode prompt registration
- entry-like
zim_file_pathfallback in single-archive mode - reduced prompt surface
- Wikipedia-oriented simple-mode prompts over
zim_query
License
This repository remains under the existing project license unless you change it separately.
Recommended Servers
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.
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.
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.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
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.
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.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
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.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.