OpenClaw

OpenClaw can manage outbound MCP server definitions and expose those tools to supported runtimes. Add Iteration Layer as a remote Streamable HTTP MCP server to give OpenClaw agents a complete content-processing toolkit.

Use this when an OpenClaw agent needs to extract data with Document Extraction, convert files with Document to Markdown, process images with Image Transformation, or generate client-ready outputs with Document Generation.

Requirements

• OpenClaw with MCP support
• An Iteration Layer account
• Browser access for the first OAuth authorization if your runtime initiates OAuth interactively

The Iteration Layer MCP server uses OAuth 2.1. Prefer OAuth when your OpenClaw runtime supports it. Do not hard-code Iteration Layer API keys into MCP headers.

MCP Server Endpoint

https://api.iterationlayer.com/mcp

Add Iteration Layer With The OpenClaw CLI

OpenClaw can save MCP server definitions through openclaw mcp set. Use Streamable HTTP transport:

openclaw mcp set iterationlayer '{
  "url": "https://api.iterationlayer.com/mcp",
  "transport": "streamable-http"
}'

Then confirm the server definition is stored:

openclaw mcp show iterationlayer

OpenClaw stores this under the top-level mcp.servers config block for runtimes that consume OpenClaw-managed MCP definitions.

Add Iteration Layer In Config

If you manage OpenClaw config directly, add a remote MCP server definition under mcp.servers:

{
  mcp: {
    servers: {
      iterationlayer: {
        url: "https://api.iterationlayer.com/mcp",
        transport: "streamable-http"
      }
    }
  }
}

Some OpenClaw CLI paths accept type: "http" and normalize it to transport: "streamable-http". Use transport in checked-in config because it is the canonical shape for embedded runtimes.

Verify The Connection

Start the OpenClaw runtime that consumes your configured MCP servers, then ask the agent:

List the Iteration Layer tools you can use.

You should see tools such as:

• convert_document_to_markdown
• extract_document
• extract_website
• transform_image
• generate_image
• generate_document
• generate_sheet

First Workflow

Use a small, inspectable file before connecting OpenClaw to larger project data:

Convert this project brief to markdown, extract client name, deliverables, due dates, and open questions, then generate a client kickoff summary as a PDF.

The agent can call convert_document_to_markdown for context, extract_document for typed fields, and generate_document for the final output.

Good OpenClaw Workflows

OpenClaw works well when an agent owns an operational task and needs external tools with predictable inputs and outputs.

Common workflows:

• Client document intake and summary generation
• Supplier invoice extraction and approval report creation
• Product catalog parsing, image preparation, and listing sheet generation
• Batch document review where low-confidence fields need human follow-up

Production Guidance

Use OpenClaw and MCP for agent-operated work, workflow exploration, and internal operations. For product flows that must run deterministically, call Iteration Layer through REST or an SDK.

The same content capabilities are available through MCP and REST, so you can prototype the workflow with OpenClaw and move the stable parts into production code later.

Troubleshooting

OpenClaw lists the server but the agent does not see tools

The openclaw mcp set command stores the server definition. The runtime still needs to consume OpenClaw-managed MCP definitions. Restart the runtime and confirm it supports remote MCP tool discovery.

Remote connection fails

Confirm the server URL is https://api.iterationlayer.com/mcp and the transport is streamable-http.

OAuth is not available in your runtime

Use an OpenClaw runtime that supports OAuth for remote MCP servers. Iteration Layer MCP is designed for OAuth 2.1 rather than static API-key headers.

Tool calls time out

Increase the remote MCP connection timeout for large documents or generation-heavy workflows.