Cortex Cloud Docs

You can search Cortex Cloud documentation and API references through a dedicated MCP server that exposes indexing and search functionality. This server lets Claude and other MCP clients quickly locate relevant information, improving how you explore Cortex Cloud docs and API references.

How to use

Set up your MCP client to connect to the Cortex Cloud Docs server. You will index Cortex Cloud documentation and API references first, then perform searches to locate precise information across both sources. Use the server to run targeted queries like product docs, API references, and related concepts, and rely on caching and real-time indexing to speed up repeated searches.

Before searching, index the documentation you need: run the indexing tools to populate the local cache. Once indexing completes, you can perform queries that span Cortex Cloud documentation and its API docs. If you need to refresh results, trigger indexing again to keep the cache up to date.

How to install

Prerequisites: Python 3.12 or higher and the uv package manager.

Option 1: From PyPI (Recommended) Install is not required. You will reference the server via uvx in your Claude Desktop configuration.

# Option 1: PyPI (Recommended) – configure in Claude Desktop
{
  "mcpServers": {
    "Cortex Cloud Docs": {
      "command": "uvx",
      "args": ["cortex-cloud-docs-mcp-server@latest"],
      "env": {},
      "transport": "stdio"
    }
  }
}

Option 2: Development Installation Follow these steps to run locally. Prerequisites are Python 3.12+ and uv.

# Prerequisites
curl -LsSf https://astral.sh/uv/install.sh | sh

# Clone the project and set up
git clone https://github.com/clarkemn/cortex-cloud-docs-mcp-server.git
cd cortex-cloud-docs-mcp-server
uv sync

Additional setup and testing

If you are testing manually, you can initiate a test session using the MCP client protocol by sending an initialize message and then issuing search/lookup requests after indexing. Use the appropriate client tooling to pass through the initialization and query phases, ensuring the server has completed its indexing tasks before querying.

Development focus: You can run the server locally with the command shown in the development flow and verify its behavior against the test payloads.

Notes and troubleshooting

Indexing must complete before searches to ensure results are available. If you encounter issues starting the server in Claude Desktop, verify that uv is installed and the path to the project directory is correct. Check client logs for specific error messages and rerun the indexing steps if necessary.

Available tools

index_cortex_docs

Index Cortex Cloud documentation to make it searchable by the MCP server.

index_cortex_api_docs

Index Cortex Cloud API documentation to extend search scope.

search_cortex_docs

Search the Cortex Cloud documentation index for user queries.

search_cortex_api_docs

Search the Cortex Cloud API documentation index for user queries.

search_all_docs

Search across all indexed Cortex Cloud documentation and API docs.

get_index_status

Check indexing progress and cache statistics to monitor server health.