TurboLynx Installation

Client

Command line

C / C++

Python

Java

Node.js

MCP

Platform

Linux

macOS

Windows

Installation

git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx
cmake -GNinja -DCMAKE_BUILD_TYPE=Release \
      -DENABLE_TCMALLOC=OFF -DBUILD_UNITTESTS=OFF -DTBB_TEST=OFF \
      -B build
cmake --build build

Verify

./build/tools/turbolynx --help

git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx
cmake -GNinja -DCMAKE_BUILD_TYPE=Release \
      -DENABLE_TCMALLOC=OFF -DBUILD_UNITTESTS=OFF -DTBB_TEST=OFF \
      -B build
cmake --build build

This produces the shared library at ${TURBOLYNX_DIR}/build/src/libturbolynx.so and exposes the public C header at ${TURBOLYNX_DIR}/src/include/main/capi/turbolynx.h.

Link from your application

target_include_directories(your_app PRIVATE
    ${TURBOLYNX_DIR}/src/include)
target_link_libraries(your_app PRIVATE
    ${TURBOLYNX_DIR}/build/src/libturbolynx.so)

#include "main/capi/turbolynx.h"

int64_t conn = turbolynx_connect("/path/to/workspace");
if (conn < 0) { /* handle error */ }
/* ... use conn ... */
turbolynx_disconnect(conn);

See the C API reference for the full surface.

Build the native binary on the Mac you plan to run it on. Do not download a Linux build/ directory or wheel and reuse it on macOS.

xcode-select --install
brew install cmake ninja
git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx
cmake -GNinja -DCMAKE_BUILD_TYPE=Release \
      -DTURBOLYNX_PORTABLE_DISK_IO=ON -DENABLE_TCMALLOC=OFF \
      -DBUILD_UNITTESTS=OFF -DTBB_TEST=OFF \
      -B build-portable
cmake --build build-portable

Verify

./build-portable/tools/turbolynx --help

Windows is not supported.

Build the shared library on the target Mac. Native artifacts are platform-specific.

xcode-select --install
brew install cmake ninja
git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx
cmake -GNinja -DCMAKE_BUILD_TYPE=Release \
      -DTURBOLYNX_PORTABLE_DISK_IO=ON -DENABLE_TCMALLOC=OFF \
      -DBUILD_UNITTESTS=OFF -DTBB_TEST=OFF \
      -B build-portable
cmake --build build-portable

This produces the shared library at ${TURBOLYNX_DIR}/build-portable/src/libturbolynx.dylib and exposes the public C header at ${TURBOLYNX_DIR}/src/include/main/capi/turbolynx.h.

Link from your application

target_include_directories(your_app PRIVATE
    ${TURBOLYNX_DIR}/src/include)
target_link_libraries(your_app PRIVATE
    ${TURBOLYNX_DIR}/build-portable/src/libturbolynx.dylib)

#include "main/capi/turbolynx.h"

int64_t conn = turbolynx_connect("/path/to/workspace");
if (conn < 0) { /* handle error */ }
/* ... use conn ... */
turbolynx_disconnect(conn);

See the C API reference for the full surface.

Windows is not supported.

Build TurboLynx, then build and install the Python wheel from tools/pythonpkg/:

git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx
python3 -m pip install pybind11 wheel
cmake -GNinja -DCMAKE_BUILD_TYPE=Release -DBUILD_PYTHON=ON \
      -DENABLE_TCMALLOC=OFF -DBUILD_UNITTESTS=OFF -DTBB_TEST=OFF \
      -B build
cmake --build build
tools/pythonpkg/scripts/build_wheel.sh build
python3 -m pip install tools/pythonpkg/dist/turbolynx-*.whl

Verify

python3 -c 'import turbolynx; print(turbolynx.__version__)'

A pre-built pip install turbolynx wheel on PyPI is on the roadmap. See the Python API reference for usage.

Build and install the wheel on the same Mac and architecture that will import it. A Linux wheel cannot be reused on macOS.

xcode-select --install
brew install cmake ninja
python3 -m pip install pybind11 wheel
git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx
cmake -GNinja -DCMAKE_BUILD_TYPE=Release -DBUILD_PYTHON=ON \
      -DTURBOLYNX_PORTABLE_DISK_IO=ON -DENABLE_TCMALLOC=OFF \
      -DBUILD_UNITTESTS=OFF -DTBB_TEST=OFF \
      -B build-portable
cmake --build build-portable
tools/pythonpkg/scripts/build_wheel.sh build-portable
python3 -m pip install tools/pythonpkg/dist/turbolynx-*.whl

Verify

python3 -c 'import turbolynx; print(turbolynx.__version__)'

Python bindings are on the roadmap.

JNI bindings are on the roadmap.

The Node.js package wraps the TurboLynx WebAssembly runtime, so it works without any native build step. From a checkout:

git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx/tools/nodepkg
npm install
npm pack   # produces turbolynx-0.0.1.tgz
npm install ./turbolynx-0.0.1.tgz   # install into your project

Verify

node -e "const {TurboLynx} = require('turbolynx'); TurboLynx.version().then(v => console.log(v))"

A pre-built npm install turbolynx on the npm registry is on the roadmap.

The MCP server plugs TurboLynx into Claude Desktop, Cursor, Codex, and any other Model Context Protocol client. It bundles the WASM runtime, so no native build is needed.

git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx/tools/mcp
npm install
npm pack
npm install -g ./turbolynx-mcp-0.0.1.tgz

Claude Desktop config

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop:

{
  "mcpServers": {
    "turbolynx": {
      "command": "turbolynx-mcp",
      "env": { "TURBOLYNX_WORKSPACE": "/path/to/workspace" }
    }
  }
}

See the MCP server reference for the tool / resource catalog.

The MCP server installs the same way on macOS because it wraps the WASM-based Node.js runtime. If you need to create a workspace locally, build it on macOS with the CLI, C API, or Python instructions above. Otherwise point TURBOLYNX_WORKSPACE at an existing workspace directory.

git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx/tools/mcp
npm install
npm pack
npm install -g ./turbolynx-mcp-0.0.1.tgz

Claude Desktop config

Edit ~/Library/Application Support/Claude/claude_desktop_config.json, then restart Claude Desktop:

{
  "mcpServers": {
    "turbolynx": {
      "command": "turbolynx-mcp",
      "env": {
        "TURBOLYNX_WORKSPACE": "/path/to/workspace"
      }
    }
  }
}

See the MCP server reference for the tool and resource catalog.

The MCP package itself installs on Windows, but native TurboLynx workspace creation is not documented yet. Point it at an existing workspace directory.

The Node.js package wraps the TurboLynx WebAssembly runtime, so installation is the same on macOS and Linux.

git clone https://github.com/turbolynx-dslab/TurboLynx
cd TurboLynx/tools/nodepkg
npm install
npm pack
npm install ./turbolynx-0.0.1.tgz

Verify

node -e "const {TurboLynx} = require('turbolynx'); TurboLynx.version().then(v => console.log(v))"

The Node.js package itself installs on Windows, but native TurboLynx workspace creation is not documented yet. Point it at an existing workspace directory.