MCP Server for Claude: ToolYour Agent Setup

Integrating a robust set of utilities into your Claude Desktop or Claude Code workflow empowers your AI agent to move beyond basic coding tasks and tackle complex real-world challenges. ToolYour provides a powerful remote Model Context Protocol (MCP) server that extends Claude's capabilities with a wide array of API-backed tools. This guide will walk you through the essential steps to configure your Claude client to utilize the ToolYour MCP server, leverage the agent-centric solve_task function, and understand how billing and quotas are shared with our REST API.

Whether your agent needs to perform a full SEO audit, run security checks, convert documents, or manage other business-critical tasks, ToolYour's MCP server for Claude offers a streamlined plan → run → verify workflow. By the end of this setup guide, your Claude agent will be ready to execute intricate jobs with precision and efficiency.

Why Claude Needs a Utilities MCP Server

Claude is an exceptionally capable conversational AI, adept at understanding context, generating code, and assisting with intricate reasoning. However, to truly act as an autonomous agent in dynamic environments, Claude needs access to external utilities. These tools allow it to interact with web services, process files, perform calculations, and conduct specialized analyses that go beyond its core language model capabilities.

A remote MCP server like ToolYour bridges this gap, providing Claude with a catalog of API-backed tools designed to complete jobs such as ship-gate validation, comprehensive SEO audits, and thorough security analyses. The Model Context Protocol (https://modelcontextprotocol.io/introduction) provides a standardized way for AI agents to discover, describe, and execute functions from external services. ToolYour implements this protocol to ensure seamless integration, enabling your Claude agent to plan its actions, execute complex playbooks or solve specific goals, and verify the outcomes until the task is successfully completed. This allows your agent to automate workflows that previously required manual intervention or custom integrations.

Get a ToolYour API Key

To connect your Claude client to the ToolYour MCP server, you first need a ToolYour API key. This key authenticates your requests and links them to your account's usage quota.

Follow these steps to obtain your API key:

1. Navigate to the ToolYour dashboard: https://www.toolyour.com/dashboard.
2. If you don't have an account, sign up at https://www.toolyour.com/signup.
3. Once logged in, go to the API Keys section: https://www.toolyour.com/dashboard/api-keys.
4. Generate a new API key. It will start with the prefix ty_. Keep this key secure, as it grants access to your ToolYour services.

The API key you generate provides access to all ToolYour services, including both the REST API and the MCP server. Your monthly credit quota and associated usage limits are shared across all services and any API keys linked to your account. The Free plan offers 500 credits per month, with tools costing 1–10 credits per successful call. Additionally, the Free plan allows for 30 successful tool runs within any rolling 60-second window. Quota resets on the 1st of each calendar month.

Claude Desktop Configuration

Configuring Claude Desktop to use the ToolYour MCP server involves modifying its mcp.json configuration file, which defines the external MCP servers your agent can access.

1. Locate mcp.json: The mcp.json file is typically found in your Claude Desktop application's configuration directory. The exact location can vary by operating system and installation, but common paths include directories within your user profile or application support folders. Refer to the official Claude documentation for the precise location if needed (https://docs.anthropic.com/en/docs/agents-and-tools/mcp).

2. Add ToolYour Server Configuration: Open your mcp.json file with a text editor. You will add a new entry under the mcpServers object for ToolYour. Ensure your mcp.json structure is valid JSON after the addition.

Here is the required JSON snippet for the ToolYour MCP server configuration:

{
  "mcpServers": {
    "toolyour": {
      "url": "https://api.toolyour.com/mcp",
      "headers": {
        "X-Api-Key": "ty_YOUR_KEY"
      }
    }
  }
}

Important: Replace ty_YOUR_KEY with the actual API key you obtained from https://www.toolyour.com/dashboard/api-keys.

• url: This specifies the endpoint for the ToolYour MCP server. All MCP requests will be directed here.
• headers: This section contains the HTTP headers that will be sent with every request to the ToolYour MCP server. The X-Api-Key header is crucial for authentication.

3. Save and Restart: Save the mcp.json file and restart your Claude Desktop application. Upon restart, Claude Desktop will load the new configuration and make the ToolYour MCP server available to your agent.

For a comprehensive guide on general MCP client setup, including troubleshooting common configuration issues, visit our developer documentation at https://www.toolyour.com/developers/mcp. You can also find more details on configuring MCP in Claude's official documentation (https://docs.anthropic.com/en/docs/agents-and-tools/mcp).

Claude Code Configuration Differences

When working with Claude Code, the principles for integrating the ToolYour MCP server remain largely the same as with Claude Desktop. The primary difference often lies in where and how the mcp.json or equivalent configuration is managed within your development environment or project structure.

In Claude Code, your mcp.json might be located directly within your project's root directory or a designated configuration folder, making it part of your version-controlled repository. Some setups might also leverage environment variables for sensitive information like API keys, though the mcp.json approach with direct header injection is common and supported.

The JSON structure provided for Claude Desktop configuration is identical for Claude Code:

{
  "mcpServers": {
    "toolyour": {
      "url": "https://api.toolyour.com/mcp",
      "headers": {
        "X-Api-Key": "ty_YOUR_KEY"
      }
    }
  }
}

Remember to replace ty_YOUR_KEY with your actual ToolYour API key. After modifying or creating the mcp.json file in your Claude Code project, ensure your environment is reloaded or your application restarted for the changes to take effect. Always consult the specific documentation for your Claude Code setup if you encounter configuration-related issues, or refer to our general MCP setup guide at https://www.toolyour.com/developers/mcp.

Using solve_task from a Conversation

The solve_task meta-tool is the primary entry point for agents to achieve complex goals on the ToolYour MCP server. It allows your Claude agent to articulate a plain-language objective, which ToolYour then routes to appropriate workflows or tools using fuzzy matching and confidence gating.

The typical agent loop when interacting with ToolYour's MCP server is:

1. plan_task (Free): The agent initiates a plan to understand the scope and potential tools for a given goal. This step does not consume credits.

2. solve_task or run_playbook: Based on the plan, the agent then calls solve_task with a specific goal or run_playbook for a predefined sequence of actions. For example, your Claude agent might send a request like this to address a document conversion need:

   {
     "goal": "Convert report.docx to PDF",
     "input": {
       "url": "https://yoursite.com/report.docx"
     },
     "responseMode": "compact"
   }
   

   ToolYour prioritizes a payloadFirstRule, meaning it's best to pass workspace files, HTML, or text directly when possible. Only provide a URL if the user explicitly requested live link analysis, or if the matched job cannot run without fetching external content. If the goal is ambiguous, solve_task will return a status: suggest response with ranked suggestions, allowing the agent to clarify its intent rather than executing an incorrect operation.

3. verify_task: After solve_task or run_playbook execution, the agent uses verify_task to check if the goal was successfully achieved. This meta-tool evaluates the outcome against a baseline, providing deltas on status, scores, new findings, and remaining fixes. The agent should apply the top-ranked item from loop.nextActions (which is part of loop.remainingFixes) and re-run verify_task until loop.gate indicates pass.

It is important to note that while plan_task and catalog exploration are free, executing solve_task or run_playbook to achieve a goal does consume credits from your monthly allowance, similar to REST API calls. Each successful tool call within these operations costs between 1 and 10 credits. For a more detailed understanding of solve_task workflows, our developers section provides in-depth articles.

While discover_tools, get_tool_schema, and invoke_tool are available for explicit, one-off operations, solve_task and run_playbook are the recommended paths for agent-driven job completion due to their robustness and goal-oriented design.

API-Backed Tools Only

ToolYour offers a comprehensive catalog of over 200 tools, accessible through its browser interface, REST API, and MCP server. However, it's crucial for developers to understand a key distinction when integrating with the MCP server: ToolYour's MCP exposes API-backed tools only.

This design decision ensures that agents interacting with the MCP server can reliably automate tasks without requiring a browser environment or manual intervention. Tools that are exclusively designed for browser-only use cases, often relying on interactive UI elements or client-side JavaScript, are not exposed via the MCP endpoint.

Therefore, while the ToolYour https://www.toolyour.com/ai-tools catalog might list a wide range of utilities for various purposes—from https://www.toolyour.com/developer-tools/json-validator to https://www.toolyour.com/file-conversion/docx-to-pdf-converter—your Claude agent will only be able to access those that have a corresponding API implementation through the https://api.toolyour.com/mcp endpoint. This focused approach provides a stable and predictable environment for agent automation, allowing you to build robust workflows for SEO, security, and ship-gate tasks.

Shared Quota with REST

One of the significant advantages of the ToolYour platform is its unified billing system across all interfaces. Your ToolYour API key, generated from https://www.toolyour.com/dashboard/api-keys, grants access to both the REST API and the MCP server, and critically, both share the same monthly credit quota.

This means that whether your Claude agent is making calls to https://api.toolyour.com/mcp or you are directly interacting with our REST API endpoints (e.g., https://api.toolyour.com/api/v1/documents/docx-to-pdf), all successful tool executions draw from the same allowance.

Let's break down the quota details:

• Free Plan: Includes 500 credits per month. Tools generally cost between 1 and 10 credits for each successful call.
• Throughput Limit: In addition to monthly credits, the Free plan also has a throughput limit of 30 successful tool runs per rolling 60-second window.
• Quota Reset: Your credit quota resets on the 1st of each calendar month.
• Multiple API Keys: If you generate multiple API keys under one ToolYour account, they all share the same overarching monthly credit quota.
• Credit Consumption: Only successful tool calls consume credits. Invalid requests or discovery operations like browsing the MCP catalog or using plan_task (without execution) are free and do not deduct from your allowance.

This shared quota system simplifies usage management, allowing you to flexibly use ToolYour services across different automation needs without worrying about separate billing for MCP or REST. For more information on plans and credit allocations, visit https://www.toolyour.com/pricing. If your monthly credit quota or throughput limit is reached, further requests will be blocked until your plan is upgraded or until the next calendar month's reset.

FAQ

Q: What is the Model Context Protocol (MCP)? A: The Model Context Protocol (MCP) is a standard that allows AI agents to discover, understand, and interact with external tools and services. ToolYour implements MCP to provide agents like Claude with access to its API-backed utility catalog (https://modelcontextprotocol.io).

Q: Do I need a separate API key for ToolYour's MCP and REST API? A: No, you use the same ToolYour API key (starting with ty_) for both the MCP server and the REST API. Your monthly credit quota is shared across all services accessed with this key.

Q: How many tools are available through ToolYour's MCP server? A: ToolYour's full catalog includes over 200 tools. The MCP server specifically exposes the subset of these tools that are API-backed, ensuring reliable automation for your agent workflows.

Q: What happens if my Claude agent exceeds my monthly ToolYour credit quota? A: If your monthly credit quota or the rolling 60-second throughput limit is reached, subsequent requests from your Claude agent to the ToolYour MCP server will be blocked. To continue usage, you will need to upgrade your plan or wait until your quota resets on the 1st of the next calendar month.

Q: Is using solve_task always free on ToolYour's MCP? A: No. While the plan_task meta-tool and receiving suggestions for ambiguous goals are free, the execution of a solve_task operation to achieve a specific goal does consume credits from your monthly allowance, just like REST API calls.

Q: Can I use ToolYour MCP for ship-gate or SEO tasks with Claude? A: Absolutely. ToolYour's remote MCP server is designed specifically to help agents complete complex ship-gate, SEO, security, and other utility-oriented jobs through a structured plan → run → verify loop.

Q: Where can I find examples of specific jobs or playbooks available through ToolYour MCP? A: ToolYour's MCP supports a variety of job workflows and playbooks, including "full-seo-audit," "security-headers-job," and "document-convert-pipeline." You can explore examples and more details in our developer documentation at https://www.toolyour.com/developers/mcp.

Conclusion

By following this setup guide, you have successfully configured your Claude Desktop or Claude Code client to leverage the powerful ToolYour MCP server. Your agent is now equipped to execute a wide range of API-backed utilities, from enhancing SEO performance and running security audits to automating ship-gate checks and document conversions.

The integration of ToolYour's remote MCP server allows your Claude agent to operate with greater autonomy and effectiveness in real-world scenarios. Remember the unified credit quota across REST and MCP, and the agent-centric workflow facilitated by solve_task, plan_task, and verify_task.

Take your agent's capabilities to the next level:

• Get your API key: If you haven't already, secure your key at https://www.toolyour.com/dashboard/api-keys.
• Explore ToolYour's capabilities: Discover more tools and workflows at https://www.toolyour.com/ai-tools or https://www.toolyour.com/developers/mcp.
• Review pricing plans: Understand credit allocations and throughput limits at https://www.toolyour.com/pricing.