Skip to main content
Claude is an AI platform built by Anthropic. Claude excels at AI-driven tasks involving language, reasoning, analysis, coding, and more. Claude Desktop gives you a graphical interface built for running multiple Claude sessions side by side. Claude Desktop provides access to three modes of working:
  • Claude Chat, which provides general conversation with no file access, similar to claude.ai. At the time this article was initially published, Claude Chat was available on all Claude pricing plans.
  • Claude Cowork, an autonomous background agent that works on tasks in a cloud virtual machine with its own environment. It can run independently while you do other work. At the time this article was initially published, Claude Cowork was available only on Claude Pro, Max, Team, and Enterprise pricing plans.
  • Claude Code, an interactive coding assistant with direct access to your local files. You review and approve each change in real time. At the time this article was initially published, Claude Code was available only on Claude Pro, Max, Team, and Enterprise pricing plans.
This article shows how you can use the Claude Cowork interface within Claude Desktop to interact with Unstructured document elements and metadata that have been sent to various destinations that Unstructured supports. For example, you could send through Unstructured your organization’s source documents stored for instance in an Amazon S3 bucket that you own, which transforms those documents’ contents into AI-ready data. You could then have Unstructured send its outputs for instance into a DataStax Astra database that you own. You could then connect Claude Desktop to this database and, finally, use Claude Cowork to interact with the data in this database.

​
Requirements

Before you can use the Claude Cowork interface within Claude Desktop to interact with your destination location that contains Unstructured outputs, you must do the following:
  1. Sign up for a Claude paid plan or upgrade from a free to a paid Claude plan, if you have not already done so.
  2. Install Claude Desktop on your local machine, if you have not already done so.
  3. Create an Unstructured workflow that connects one or more supported sources containing your organization’s source documents, semi-structured data, or both, to a supported destination that will contain Unstructured’s AI-ready outputs based on those source documents, semi-structured data, or both. This article has been tested with the following Unstructured destination types:
    • Astra DB
    • Google Drive
    • S3
    This list will be updated over time as additional Unstructured destination types are tested. Any supported Unstructured destination type that has a corresponding available Model Context Protocol (MCP) server should work. Most of these MCP servers are available through third parties and not through Unstructured.
    MCP servers that are offered by third parties are neither owned nor supported by Unstructured. For questions, issues, or feature requests for a specific MCP server, contact that MCP server’s provider.
  4. Enhance your workflow output’s relevance, context, and insights by adding enrichments as desired.
  5. Add chunking and embedding to your workflow as desired.
    While Claude Cowork can handle large files and extensive context, you do not need to manually chunk data first, but doing so can lead to better performance and efficiency for very large tasks.For very large-scale use cases involving millions of Unstructured outputs, embeddings are recommended for Claude Cowork, but they are not required for standard, file-based tasks.
  6. Run your workflow as a job. This makes Unstructured generate AI-ready outputs based on your source documents, semi-structured data, or both, and store its outputs in your destination.
  7. Monitor your workflow’s job run. After your job has successfully completed, continue by finding the following section that matches your Unstructured destination type and completing that section’s instructions, to connect Claude Desktop to your destination.
This article has been tested with the following Unstructured destination types: This list will be updated over time as additional Unstructured destination types are tested.

​
Astra DB

This section references a third-party MCP server that is neither owned nor supported by Unstructured. For technical support, questions, issues, or feature requests for this MCP server, contact the MCP server’s provider.For more information, see the Astra DB MCP Server repository in GitHub.
After you have completed the requirements, do the following:
  1. In the README file in the Astra DB MCP Server repository in GitHub, follow the instructions in these sections, in this order:
    1. Prerequisites
    2. Claude Desktop
  2. Start Claude Desktop. If it is already running, stop and then restart it. This step helps to ensure that Claude Desktop can successfully start and use the MCP server.
  3. Sign in to Claude by following Claude Desktop’s on-screen instructions.
  4. Switch to Claude Cowork mode: in the upper-left corner of Claude Desktop, click the Cowork icon.
  5. In the How can I help you today box, enter your question about the Unstructured outputs within your target Astra DB collection, and then press Enter. Repeat this step as needed.
    To constrain Claude Cowork to use only this MCP server, you should be explicit in your sessions’ initial prompt or within your Claude project’s files such as CLAUDE.md, for example:
    Use ONLY the Astra DB collection "<your-collection-name>" as your source of truth.

Rules:

- For every question, retrieve relevant records using the
  mcp__astra-db-mcp__VectorSearch or HybridSearch tool against that
  collection before answering.
- Do NOT use web search, WebFetch, your own training knowledge, or
  any other MCP connector. Do NOT guess or fill in gaps from memory.
- Base every factual claim on retrieved records. Quote or paraphrase
  the retrieved `content` field and cite the source `filename` and
  `page_number` from each record's metadata.
- If the collection does not contain enough information to answer,
  say exactly: "The collection does not contain information on this."
  Do not speculate or supplement from other sources.
- If I ask something that requires tools other than Astra DB (such as 
  file creation or calculations on retrieved data), you may use those
  tools, but the underlying facts must still come from the collection.

​
Google Drive

This section references a third-party MCP server that is neither owned nor supported by Unstructured. For technical support, questions, issues, or feature requests for this MCP server, contact the MCP server’s provider.For more information, see the Google Drive MCP Server repository in GitHub.
After you have completed the requirements, do the following:
  1. In the README file in the Google Drive MCP Server repository in GitHub, follow the instructions in these sections, in this order:
    1. Requirements
    2. Google Cloud Setup
    3. Configuration
    4. Usage with Claude Desktop
  2. Start Claude Desktop. If it is already running, stop and then restart it. This step helps to ensure that Claude Desktop can successfully start and use the MCP server.
  3. Sign in to Claude by following Claude Desktop’s on-screen instructions.
  4. Switch to Claude Cowork mode: in the upper-left corner of Claude Desktop, click the Cowork icon.
  5. In the How can I help you today box, enter your question about the Unstructured outputs within your Google Drive, and then press Enter. Repeat this step as needed.
    To constrain Claude Cowork to use only this MCP server, you should be explicit in your sessions’ initial prompt or within your Claude project’s files such as CLAUDE.md, for example:
    For this entire task, use only the `google-drive` MCP server to answer my questions.

Rules:

- Do NOT use web search, WebFetch, your own training knowledge, or
  any other MCP connector. Do NOT guess or fill in gaps from memory.
- Base every factual claim on retrieved records. Quote or paraphrase
  the retrieved content and cite the source `filename` and
  `page_number` from each record's metadata.
- If the files do not contain enough information to answer,
  say exactly: "The source files do not contain information on this."
  Do not speculate or supplement from other sources.
- If I ask something that requires tools other than `google-drive`, you may use those
  tools, but the underlying facts must still come from the source files.

​
S3

This section references a third-party MCP server that is neither owned nor supported by Unstructured. For technical support, questions, issues, or feature requests for this MCP server, contact the MCP server’s provider.For more information, see the S3 MCP Server repository in GitHub.
After you have completed the requirements, do the following:
  1. Start Claude Desktop.
  2. Open your Claude Desktop config file: in the lower-left corner of Claude Desktop, click your login icon, and then click Settings.
  3. In the sidebar, under Desktop app, click Developer.
  4. Click Edit Config.
  5. Follow the on-screen instructions to open your claude_desktop_config.json file in your favorite text or code editor.
  6. In the file’s mcpServers section, add the following code. (The ellipses indicate omitted code for brevity.) Replace the following placeholders, and then save and close the file:
    • Replace <aws-region> with the short ID of the AWS Region for the target S3 buckets (for example, us-east-1).
    • Replace <s3-bucket> with the name of each target S3 bucket, one entry per target bucket, with multiple bucket names separated by commas.
      Specify only bucket names here. Do not specify paths within buckets. To restrict Claude to use specific paths within buckets, specify those paths as a constraint within your prompts or Claude project files.
    • Replace <s3-max-buckets> with the number of specified buckets.
    • Replace <your-aws-access-key-id> with an AWS access key ID that has access to the buckets’ contents.
    • Replace <your-aws-secret-access-key> with the AWS secret access key for the preceding AWS access key ID. 
    {
  "mcpServers": {
    "...": "...",
    "aws-s3-mcp": {
      "command": "npx",
      "args": ["aws-s3-mcp", "--stdio"],
      "env": {
        "AWS_REGION": "<aws-region>",
        "S3_BUCKETS": "<s3-bucket-1>,<s3-bucket-2>,<s3-bucket-N>",
        "S3_MAX_BUCKETS": "<s3-max-buckets>",
        "AWS_ACCESS_KEY_ID": "<your-aws-access-key-id>",
        "AWS_SECRET_ACCESS_KEY": "<your-aws-secret-access-key>"
      }
    },
    "...": "..."
  },
  "...": "..."
}
  7. Restart Claude Desktop. This step helps to ensure that Claude Desktop can successfully start and use the MCP server.
  8. Sign in to Claude by following Claude Desktop’s on-screen instructions.
  9. Switch to Claude Cowork mode: in the upper-left corner of Claude Desktop, click the Cowork icon.
  10. In the How can I help you today box, enter your question about the Unstructured outputs within your target S3 bucket, and then press Enter. Repeat this step as needed.
    To constrain Claude Cowork to use only this MCP server, you should be explicit in your sessions’ initial prompt or within your Claude project’s files such as CLAUDE.md, for example:
    For this entire task, use only the `aws-s3-mcp` MCP server to answer my questions. 

Rules:

- Do NOT use web search, WebFetch, your own training knowledge, or
  any other MCP connector. Do NOT guess or fill in gaps from memory.
- Base every factual claim on retrieved records. Quote or paraphrase
  the retrieved content and cite the source `filename` and
  `page_number` from each record's metadata.
- If the files do not contain enough information to answer,
  say exactly: "The source files do not contain information on this."
  Do not speculate or supplement from other sources.
- If I ask something that requires tools other than `aws-s3-mcp`, you may use those
  tools, but the underlying facts must still come from the source files.