A Model Context Protocol (MCP) server that provides seamless integration between Claude and Coda.io, enabling AI-powered document automation and data manipulation.
Note: This is an unofficial MCP server developed by TJC L.P. and is not affiliated with, endorsed by, or supported by Coda. For official Coda support and documentation, please visit coda.io.
Note: Version 1.1.0+ uses
snake_casefield names (e.g.,browser_linkinstead ofbrowserLink) for Python ecosystem compatibility. See CHANGELOG for migration details if upgrading from 1.0.x.
- List documents - Search and filter your Coda docs
- Create documents - Generate new docs with optional templates
- Read document info - Get metadata about any doc
- Update documents - Modify doc properties like title and icon
- Delete documents - Remove docs (use with caution!)
- List pages - Browse all pages in a doc
- Create pages - Add new pages with rich content
- Read pages - Get page details and content
- Update pages - Modify page properties and content
- Delete pages - Remove pages from docs
- Export page content - Get full HTML/Markdown content with
begin_page_content_exportandget_page_content_export_status
- List tables - Find all tables and views in a doc
- Get table details - Access table metadata and structure
- List columns - Browse table columns with properties
- Get column info - Access column formulas and formats
- List rows - Query and filter table data
- Get specific rows - Access individual row data
- Insert/Update rows - Add or modify data with
upsert_rows - Update single row - Modify specific row data
- Delete rows - Remove single or multiple rows
- Push buttons - Trigger button columns in tables
- List formulas - Find all named formulas in a doc
- Get formula details - Access formula expressions
- Who am I - Get current user information
- Coda API Key: Get your API token from Coda Account Settings
- Python 3.11+ (including 3.14): Required for the MCP server
The Coda MCP server is available on PyPI and can be installed directly using uvx (recommended) or pip:
# Using uvx (no installation needed, just run)
uvx coda-mcp-server
# Or install globally with pip
pip install coda-mcp-server-
Clone the repository
git clone https://github.com/TJC-LP/coda-mcp-server.git cd coda-mcp-server -
Install dependencies
uv sync
-
Set your API key as an environment variable (see Configuration section below)
For using with Claude Code during development:
-
Set your API key as a shell environment variable:
# Add to your shell profile (~/.zshrc, ~/.bashrc, etc.) export CODA_API_KEY="your-coda-api-key-here" # Or set it for the current session export CODA_API_KEY="your-coda-api-key-here"
-
MCP configuration is already included!
The repository includes a
.mcp.jsonfile that automatically configures the Coda MCP server:{ "mcpServers": { "coda": { "command": "uv", "args": ["run", "coda-mcp-server"], "env": { "CODA_API_KEY": "${CODA_API_KEY}" } } } }The
${CODA_API_KEY}syntax reads the API key from your shell environment. -
Reload Claude Code - The MCP server will be automatically available
Security Note: API keys are read from your shell environment, not from files. This prevents accidental commits and arbitrary file loading.
Alternative: Using .env files with dotenv-cli
If you prefer file-based configuration, you can use dotenv-cli to inject environment variables:
# Create .env file from example (gitignored)
cp .env.example .env
# Edit .env and replace 'changeme' with your API key
# Run Claude Code with dotenv (no installation needed)
bunx dotenv-cli -- claude
# or
npx dotenv-cli -- claudeTo use the Coda MCP server with Claude Desktop, you need to add it to your Claude Desktop configuration file.
Configuration File Location:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Add the Coda MCP Server:
Edit the configuration file and add the Coda server to the mcpServers section:
{
"mcpServers": {
"coda": {
"command": "uvx",
"args": ["coda-mcp-server"],
"env": {
"CODA_API_KEY": "your-coda-api-key-here"
}
}
}
}Important: Replace
your-coda-api-key-herewith your actual Coda API key from Coda Account Settings.
If you installed from source, you can point to your local installation:
{
"mcpServers": {
"coda": {
"command": "uv",
"args": [
"run",
"--directory",
"/path/to/repo",
"coda-mcp-server"
],
"env": {
"CODA_API_KEY": "your-coda-api-key-here"
}
}
}
}After adding the configuration:
- Restart Claude Desktop
- Look for the settings icon in the bottom of your conversation
- Click it and verify that "coda" is listed as a connected server
- You should see 26 available Coda tools when you click
codain the dropdown and can toggle the ones that you need specifically.
Once installed, you can use Coda operations directly in Claude by prefixing commands with coda:. Here are some examples:
# List all your docs
Use coda:list_docs with is_owner: true, is_published: false, query: ""
# Get info about a specific doc
Use coda:get_doc_info with doc_id: "your-doc-id"
# Create a new doc
Use coda:create_doc with title: "My New Doc"
# List tables in a doc
Use coda:list_tables with doc_id: "your-doc-id"
# Get all rows from a table with column names
Use coda:list_rows with doc_id: "your-doc-id", table_id_or_name: "Table Name", use_column_names: true
# Insert a new row
Use coda:upsert_rows with doc_id: "your-doc-id", table_id_or_name: "Table Name", rows_data: [{
cells: [
{column: "Name", value: "John Doe"},
{column: "Email", value: "john@example.com"}
]
}]
# Start page export
Use coda:begin_page_content_export with doc_id: "your-doc-id", page_id_or_name: "Page Name", output_format: "markdown"
# Check export status and get content
Use coda:get_page_content_export_status with doc_id: "your-doc-id", page_id_or_name: "Page Name", request_id: "request-id-from-previous-step"
list_docs(is_owner, is_published, query, ...)- List available docsget_doc_info(doc_id)- Get document metadatacreate_doc(title, source_doc?, timezone?, ...)- Create new documentupdate_doc(doc_id, title?, icon_name?)- Update document propertiesdelete_doc(doc_id)- Delete a document
list_pages(doc_id, limit?, page_token?)- List pages in a docget_page(doc_id, page_id_or_name)- Get page detailscreate_page(doc_id, name, subtitle?, ...)- Create new pageupdate_page(doc_id, page_id_or_name, ...)- Update page propertiesdelete_page(doc_id, page_id_or_name)- Delete a pagebegin_page_content_export(doc_id, page_id_or_name, output_format?)- Start async page exportget_page_content_export_status(doc_id, page_id_or_name, request_id)- Poll export status and download content
list_tables(doc_id, limit?, sort_by?, ...)- List all tablesget_table(doc_id, table_id_or_name)- Get table detailslist_columns(doc_id, table_id_or_name, ...)- List table columnsget_column(doc_id, table_id_or_name, column_id_or_name)- Get column details
list_rows(doc_id, table_id_or_name, query?, ...)- List and filter rowsget_row(doc_id, table_id_or_name, row_id_or_name, ...)- Get specific rowupsert_rows(doc_id, table_id_or_name, rows_data, ...)- Insert or update rowsupdate_row(doc_id, table_id_or_name, row_id_or_name, row, ...)- Update single rowdelete_row(doc_id, table_id_or_name, row_id_or_name)- Delete single rowdelete_rows(doc_id, table_id_or_name, row_ids)- Delete multiple rowspush_button(doc_id, table_id_or_name, row_id_or_name, column_id_or_name)- Trigger button
list_formulas(doc_id, limit?, sort_by?)- List named formulasget_formula(doc_id, formula_id_or_name)- Get formula details
whoami()- Get current user information
coda-mcp-server/
├── src/
│ ├── coda_mcp_server/
│ │ ├── server.py # MCP server orchestrator (700 lines)
│ │ ├── client.py # HTTP client with Pydantic serialization
│ │ ├── models/ # 83 Pydantic models (7 modules)
│ │ │ ├── __init__.py
│ │ │ ├── common.py # Shared types and base models
│ │ │ ├── docs.py # Document models
│ │ │ ├── pages.py # Page models
│ │ │ ├── tables.py # Table and column models
│ │ │ ├── rows.py # Row and cell models
│ │ │ ├── exports.py # Export workflow models
│ │ │ └── formulas.py # Formula models
│ │ └── tools/ # Pure functions (5 modules)
│ │ ├── __init__.py
│ │ ├── docs.py # Document operations
│ │ ├── pages.py # Page operations
│ │ ├── tables.py # Table operations
│ │ ├── rows.py # Row operations
│ │ └── formulas.py # Formula operations
│ └── resources/
│ └── coda-openapi.yml # Coda API specification
├── tests/ # 44 tests
│ ├── conftest.py
│ ├── test_models.py
│ └── test_client_requests.py
├── .env.example
├── .mcp.json # Claude Code integration
└── pyproject.toml
# Install dependencies
uv sync
# Set your API key (if not already in your shell profile)
export CODA_API_KEY="your-coda-api-key-here"
# Run the server directly
uv run python src/coda_mcp_server/server.py-
"API Error 401: Unauthorized"
- Check that your
CODA_API_KEYenvironment variable is set correctly - Verify with:
echo $CODA_API_KEY - Ensure your API key has the necessary permissions
- Check that your
-
"Rate limit exceeded"
- Coda API has rate limits; wait for the specified time before retrying
- The server includes automatic rate limit detection
-
Boolean parameters not working
- The server automatically converts boolean values to strings ("true"/"false")
- This is handled internally, just use boolean values normally
-
Page export issues
- Use the two-step export workflow:
begin_page_content_exportthenget_page_content_export_status - The status check automatically downloads content when ready
- Use the two-step export workflow:
This project is licensed under the MIT License - see the LICENSE file for details.
Contributions are welcome! Please feel free to submit a Pull Request.
For issues and feature requests, please use the GitHub Issues page.