TeamSpark AI Workbench

About

TeamSpark AI Workbench is a powerful development environment for AI and machine learning projects. It is a local client
application providing a graphical interface and a command-line (terminal) interface on Mac, Linux, and Windows.

Features

TeamSpark AI Workbench supports:
- Many LLM providers and their models, including:
- Anthropic/Claude
- OpenAI/ChatGPT
- Google/Gemini
- AWS Bedrock
- Ollama
- References (memory)
- Rules (prompt guidance)
- Usage of tools via MCP (supporing thousands of available tools)
- Chat sessions where you can select and configure models, control reference and rule usage (context), and use tools.

TeamSpark AI Workbench also includes internal tools that allow models to directly interact with references and tools, meaning
the models can build and update their own references and rules (allowing them to "remember" and "learn").

CLI Mode

When building and running locally, you can launch the CLI with npm run cli

When running installed builds, see below...

MacOS

On MacOS installed releases, there is a shell script provided to launch the CLI called tspark.sh. You may run this directly,
or create a symlink to it for conveninence:

/Applications/TeamSpark\ AI\ Workbench.app/Contents/Resources/tspark.sh

or create a symlink:

ln -s /Applications/TeamSpark\ AI\ Workbench.app/Contents/Resources/tspark.sh ~/.local/bin/tspark

then just:

tspark

Linux

On Linux installed releases, TeamSpark AI Workbench is launched via teamspark-workbench. You may run in CLI mode by appending --cli.

teamspark-workbench --cli

There is also a CLI launcher called tspark.sh. You may run this directly, or create a symlink to it for convenience:

/opt/TeamSpark\ AI\ Workbench.app/tspark.sh

or create a symlink:

sudo ln -s /opt/TeamSpark\ AI\ Workbench.app/tspark.sh /usr/bin/tspark

then just:

tspark

CLI Workspace

You should either run the command line app in a directory containing a workspace, or pass it a workspace location via
the --workspace argument. To create a new workspace in current (or provided) workspace directory, use the --create
argument. Running the cli without a workspace will provide the above workspace guidance and exit.

Website

For more information about TeamSpark AI Workbench, visit our official website.

Download

Download the pre-built installer for your platform:

- macOS (Intel)
- macOS (Apple Silicon)
- Linux (Debian/Ubuntu)
- Linux (AppImage)

License

This repository is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

License Terms

The CC BY-NC-ND 4.0 license allows the work to be viewed and inspected, but it is not an open source license in the traditional sense. Specifically:

- You are prohibited from modifying the work in any way (no derivatives)
- You cannot use it for primarily commercial purposes
- This license grants the right to see and review the underlying code
- It does not permit you to adapt, build upon, or redistribute modified versions

This license may change to a more permissive license in the future if there is interest.

For commercial use licensing, please contact support@teamspark.ai.