Installation & Environment Setup
Relevant source files
The following files were used as context for generating this wiki page:
The Arclio MCP Tooling project is a monorepo containing several Model Context Protocol (MCP) packages designed for AI assistants. The installation process ranges from rapid testing using the MCP Inspector to full monorepo development setups using uv and make.
Sources: README.md:1-12, pyproject.toml:57-63
Monorepo Development Setup
The monorepo uses uv for dependency management and make for task automation. The primary setup involves cloning the repository, initializing the development environment, and installing packages in editable mode.
# Clone the repository
git clone https://github.com/arclio/arclio-mcp-tooling.git
cd arclio-mcp-tooling
# Set up the development environment
make setup-dev
# Install all packages in development mode
make install-all
Sources: README.md:43-52, pyproject.toml:16-25
Development Prerequisites
- Python 3.10 or higher: Required for all packages.
- Make: Used for executing build and development tasks.
- UV: The primary package manager used within the workspace.
Sources: README.md:38-41, pyproject.toml:9
Package-Specific Installation
Each package within the packages/ directory can be installed independently via pip or run instantly using uvx.
Quick Installation Table
| Package | PyPI Name | Primary CLI Command |
|---|---|---|
| Google Workspace MCP | google-workspace-mcp | google-workspace-worker |
| AWS S3 MCP | aws-s3-mcp | aws-s3-mcp |
| Weaviate MCP | weaviate-mcp | weaviate-worker |
| MarkdownDeck | markdowndeck | markdowndeck |
Sources: packages/google-workspace-mcp/README.md:25-45, packages/aws-s3-mcp/README.md:29-45, packages/weaviate-mcp/README.md:30-46, packages/markdowndeck/README.md:76-78
Instant Setup via MCP Inspector
For testing, packages can be run without permanent installation using the @modelcontextprotocol/inspector.
npx @modelcontextprotocol/inspector \
-e [ENVIRONMENT_VARIABLES] \
-- \
uvx --from [PACKAGE_NAME] [EXECUTABLE]
Sources: packages/google-workspace-mcp/README.md:13-23, packages/aws-s3-mcp/README.md:15-27
Environment Configuration
Configuration is primarily handled through environment variables. These can be set in the shell or via a .env file in the package or monorepo root.
Core Service Credentials
| Variable | Description | Used By |
|---|---|---|
GOOGLE_WORKSPACE_CLIENT_ID | OAuth 2.0 Client ID | Google Workspace |
GOOGLE_WORKSPACE_CLIENT_SECRET | OAuth 2.0 Client Secret | Google Workspace |
GOOGLE_WORKSPACE_REFRESH_TOKEN | OAuth 2.0 Refresh Token | Google Workspace |
AWS_ACCESS_KEY_ID | AWS access key | AWS S3 |
AWS_SECRET_ACCESS_KEY | AWS secret key | AWS S3 |
WEAVIATE_URL | Weaviate host URL | Weaviate |
WEAVIATE_OPENAI_API_KEY | OpenAI API key for vectors | Weaviate |
Sources: packages/google-workspace-mcp/README.md:105-110, packages/aws-s3-mcp/README.md:95-102, packages/weaviate-mcp/README.md:100-115
Capability Configuration
The Google Workspace MCP server uses GOOGLE_WORKSPACE_ENABLED_CAPABILITIES to control which tools are active. This must be a JSON-formatted array.
export GOOGLE_WORKSPACE_ENABLED_CAPABILITIES='["drive", "docs", "gmail", "calendar", "sheets", "slides"]'
Sources: packages/google-workspace-mcp/README.md:113-134
Authentication Flow
Different packages implement various authentication mechanisms, notably Google's OAuth 2.0 flow.
OAuth 2.0 Credential Resolution
MarkdownDeck and the Google Workspace MCP utilize a hierarchical credential lookup:
- Service Account: Checked via
GOOGLE_APPLICATION_CREDENTIALSpath. - Environment Variables: Checked for individual OAuth tokens (
SLIDES_CLIENT_ID, etc.). - Local Token Store: Checked at
~/.markdowndeck/token.json. - Interactive Flow: If no credentials exist, an
InstalledAppFlowis initiated.
Sources: packages/markdowndeck/src/markdowndeck/cli.py:32-75, packages/markdowndeck/README.md:138-146
Authentication Logic Flow
The following diagram illustrates how the system resolves credentials for Google-based services.
The diagram shows the priority-based fallback system for acquiring Google API access.
Sources: packages/markdowndeck/src/markdowndeck/cli.py:32-80
Testing the Setup
After installation, the environment can be verified using the provided Makefile targets.
# Run all tests across the monorepo
make test
# Run tests for a specific package
make test google-workspace-mcp
# Enable integration tests (requires credentials)
export RUN_INTEGRATION_TESTS="1"
make test-integration
Sources: README.md:54-68, packages/google-workspace-mcp/README.md:267-275
Summary
Setting up the Arclio MCP Tooling environment requires configuring service-specific credentials and utilizing the uv workspace for monorepo development. While individual packages can be installed via PyPI, the make setup-dev workflow provides the most robust environment for contributors and developers integrating multiple MCP services.