Agora Workbench

A workbench for wrapping your tooling with MCP

Documentation

[!IMPORTANT] Research software and support

Agora Workbench is research software and is not an officially supported Microsoft product. It is provided as-is and may change without notice. No support, service-level commitments, or compatibility guarantees are provided. Issues and contributions are welcome, but responses, fixes, and continued maintenance are not guaranteed.

Overview

Agora Workbench is a toolkit for building MCP (Model Context Protocol) servers that provide sandboxed Python execution with domain-specific packages. It is agent-framework agnostic — any MCP-compatible client can take advantage of the servers created by Agora Workbench.

Use Agora Workbench to:

• Wrap domain-specific Python tooling as MCP servers — expose Python environments through isolated, session-aware execution environments that any MCP client can call
• Make tools discoverable — register tools and skills in a searchable catalog so agents can find what they need by natural-language query
• Serve data alongside code — attach a file catalog so agents can locate and load datasets without hardcoded paths
• Deploy to Azure Container Apps — use the included Bicep templates and CLI to ship your servers with Entra ID auth and managed identity

Getting Started

Prerequisites

• Python 3.11+
• Git
• uv for dependency management
• Docker (required for the bundled container examples and local Compose workflows)

Installation

With uv (recommended for development):

git clone https://github.com/microsoft/agora-workbench.git
cd agora-workbench
uv sync --group dev  # install the project and dev tools

With pip (for using as a library):

pip install "agora-workbench==0.4.3"

The installed deployment CLI can also scaffold the standalone Activity UI:

agora-workbench-deploy init --target activity-ui

With uv, add the same release to your project with:

uv add "agora-workbench==0.4.3"

See the changelog for release history and release guide for the versioning policy.

The base install supports local code-execution servers, connectors, and local catalogs with SQLite FTS5 keyword search. It does not install Azure SDKs, sqlite-vec, or the OpenAI client.

Optional extras:

Use the command for your installation method:

# uv (inside the cloned repo)
uv sync --extra openai-agents

# pip (consuming as a library)
pip install "agora-workbench[openai-agents]==0.4.3"

For the previously bundled Azure plus hybrid-catalog feature set, install both runtime extras:

pip install "agora-workbench[azure,catalog-vector]>=0.3.0"

Installing Azure dependencies does not require cloud configuration: local no-op auth and local sources still work without credentials or endpoints. Conversely, a base-only installation means the Azure SDKs are not installed at all; selecting an Azure or vector capability reports the required extra.

The base distribution continues to include the existing execution, session, kernel, MCP, and connector dependencies. Importing the backend-neutral agora_workbench.data_lake contracts does not load execution or cloud modules. Concrete catalog, resolver, and execution integrations remain available from their documented public submodules; the extras above control only whether their cloud and vector backends are installed.

Configuration

For local testing, no external credentials are required. Use the no-op auth config to skip authentication entirely:

from agora_workbench.code_execution import CodeExecutionServer, ServerConfig
from agora_workbench.code_execution.auth import create_noop_auth_config

config = ServerConfig(
    name="myserver",
    description="My local test server",
    type="uv",
    dependency_file="numpy\npandas\n",
)
server = CodeExecutionServer(
    server_config=config,
    auth_config=create_noop_auth_config(),
)

For Docker-based local deployment and Azure Container Apps, see the deployment guide. For Entra ID authentication setup, see the authentication guide.

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit Contributor License Agreements.

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

Guidelines:

• For changes more complex than typos, please submit an issue first to discuss the proposed changes
• Follow the development practices outlined in the project documentation

Contact

For questions or feedback, please open an issue.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft’s Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party’s policies.