Choose a local setup
Windows Launcher installs Mapping, Network, and local MCP programs. It also manages the data packs and geometry packages those programs use. Start here if you want a Windows installation without setting up Python.
Local MCP server runs Geometry MCP or Research MCP against the packages in your data folder. Start here if an MCP client needs a local endpoint and you do not need the map interface.
Source checkout runs the open app, MCP server, and HTTP API from the public repository. Start here if you want to change the software or manage its deployment yourself.
Choose a data folder
Install the data packs or geometry packages your work needs in a shared data folder. Launcher manages that folder; the MCP server reads it through --data-root. Keep each extracted geometry package in its own directory with its geometry.json file.
Local programs and published downloads are free. You supply the machine and storage. Check each pack's source license before redistributing its data.
Browse content downloads or inspect pack structure and source terms before choosing a release.
Windows: install through Launcher
- Download and extract the portable Windows Launcher.
- In Settings, choose the data folder you want the programs to share.
- Install a program from Programs, then add compatible releases from Data or Geometry.
- Open the installed program from Launcher. Mapping provides the map; Geometry MCP and Research MCP provide local endpoints for an MCP client.
Program releases and data releases are separate. Check each program's detail card for supported inputs and setup notes.
Headless: make a first local MCP call
Download Geometry MCP and at least one compatible geometry package. Extract the program ZIP and place each geometry package in its own folder under your data folder. The Python program requires Python 3.11 or newer. From the extracted program folder on Windows:
py -m venv .venv
.\.venv\Scripts\python -m pip install -r requirements.txt
.\.venv\Scripts\python smoke_test.py
.\.venv\Scripts\python server.py --services geometry --data-root "D:\DaedalMapData"Use your actual data path. The smoke test checks the package offline. Once the server starts, its Geometry MCP endpoint is http://127.0.0.1:7010/mcp/geometry. A first catalog call lists the geometry available from your installed packages:
$body = '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"get_catalog","arguments":{"catalog":"geometry"}}}'
Invoke-RestMethod -Uri 'http://127.0.0.1:7010/mcp/geometry' -Method Post -ContentType 'application/json' -Headers @{'MCP-Protocol-Version'='2025-06-18'} -Body $bodyHeadless: run Research MCP locally
Download Research MCP and data packs in a folder containing catalog.json. Extract the program ZIP, install its Python requirements, and run its offline smoke test. From the extracted program folder on Windows:
py -m venv .venv
.\.venv\Scripts\python -m pip install -r requirements.txt
.\.venv\Scripts\python smoke_test.py
.\.venv\Scripts\python server.py --services research --data-root "D:\DaedalMapData"Use your actual data path. Research MCP listens at http://127.0.0.1:7011/mcp/research and queries the selected local files. Connect a Streamable HTTP MCP client to that endpoint and list its tools.
Source: run the app yourself
The public repository contains the server, MCP and HTTP API, and map frontend. Follow its README for dependencies and data installation. On Windows, clone the source and install its dependencies:
git clone https://github.com/xyver/daedal-map.git
cd daedal-map
py -m venv .venv
.\.venv\Scripts\python -m pip install -r requirements.txt
Copy-Item .env.example .envSet these values in .env, using your data folder for DATA_ROOT:
INSTALL_MODE=local
RUNTIME_MODE=local
DATA_ROOT=D:\DaedalMapData.\.venv\Scripts\python app.pyThe app opens at http://localhost:7000 and its MCP endpoint at http://localhost:7000/mcp. A blank DATA_ROOT uses the default app-data folder. Install compatible packs separately and check their source licenses before redistribution.
Hosted access
Agents and API has the remote MCP endpoint, hosted HTTP calls, access rules, and working request examples. Use that path when you want DaedalMap to run the service for your external program. Small hosted calls have free allowances; larger calls and some data queries use account credit.