MCP Server for Claude Code: ToolYour Setup

Developers using Claude Code in the terminal can integrate ToolYour's Model Context Protocol (MCP) server to access a wide range of utility tools directly within their agent workflows. This guide focuses on configuring the Claude Code MCP server for ToolYour, covering prerequisites, setup, testing, and quota considerations unique to the command-line environment.

While some core concepts of ToolYour's MCP integration are shared across various agents, this post specifically addresses the setup for Claude Code, which is typically used in a terminal or headless environment. If you're using the Claude Desktop application and prefer a UI-driven setup, refer to our dedicated post: MCP Server for Claude: ToolYour Agent Setup (Note: That article covers Claude Desktop setup details).

Why Claude Code Needs Utilities MCP

Claude Code operates within a terminal environment, and its utility often expands significantly when connected to external tools. The Model Context Protocol (MCP) provides a standardized way for AI agents like Claude Code to discover and interact with external services. ToolYour acts as a comprehensive tools platform, offering 200+ API-backed tools across categories like AI, digital, and file conversion.

Integrating ToolYour's MCP server allows your Claude Code agent to:

• Access a diverse toolkit: From ai-audio-summarizer to seo-audit, your agent gains capabilities beyond its inherent reasoning.
• Streamline workflows: Instead of manually performing tasks or calling external scripts, Claude Code can leverage ToolYour tools through direct agent interactions.
• Maintain a unified experience: ToolYour offers one catalog, one API key, and one plan that powers tools across its browser interface, REST API, and MCP for agents like Claude Code. This means your existing ToolYour API key works seamlessly.

The Model Context Protocol (MCP) is an open standard designed to enable agents to discover and use tools effectively. You can learn more about the protocol at Model Context Protocol and its application with Anthropic agents at Anthropic Agents and Tools: MCP. ToolYour provides a remote MCP server, making it easy for your Claude Code MCP server setup to connect.

API Key Prerequisites

Before configuring Claude Code for ToolYour's MCP, you need an active ToolYour API key. This key authenticates your requests and manages your usage quota.

Here’s how to get started:

1. Create a ToolYour Account: If you don't have one, sign up for a free account at ToolYour Sign Up.

2. Generate an API Key:

   • Once logged in, navigate to your API Keys dashboard: ToolYour API Keys.
   • Click "Generate New Key."
   • Your API key will start with ty_ (e.g., ty_YOUR_KEY). Keep this key secure.

Understanding Your Quota:

ToolYour's free plan provides:

• 500 successful requests per month: This quota is shared across all surfaces (REST API and MCP) and across any multiple API keys associated with your account.
• 5 successful requests per minute (throughput limit): This ensures fair usage and system stability.

Your quota resets automatically on the 1st of each calendar month. Browsing the ToolYour MCP catalog and receiving tool suggestions generally do not count towards this quota. Only actual execution of a tool through MCP consumes your allowance, similar to making a REST API call. Invalid requests that do not successfully invoke a tool typically do not count against your quota.

For detailed information on usage and plans, visit ToolYour Usage and Plans.

Claude Code MCP Configuration

Configuring Claude Code to use ToolYour's MCP server involves specifying the ToolYour MCP endpoint and your API key. Unlike a GUI-based application, Claude Code's configuration will involve editing a configuration file in your environment.

While the exact file location may vary depending on your Claude Code installation or how you manage its environment, the core configuration structure remains consistent with the Model Context Protocol specification. You will need to add a configuration block that points to the ToolYour MCP server.

Here is the JSON configuration snippet for ToolYour's MCP server:

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

Breakdown of the configuration:

• mcpServers: This is the top-level key for defining one or more MCP servers.
• toolyour: This is an arbitrary name you assign to your ToolYour MCP server. You can choose any descriptive string here.
• url: This specifies the endpoint for the ToolYour MCP server. It must be https://api.toolyour.com/mcp.
• headers: This section contains HTTP headers required for authentication.
• X-Api-Key: This is the header ToolYour uses for API key authentication.
• ty_YOUR_KEY: Replace this placeholder with your actual ToolYour API key obtained from your ToolYour Dashboard.

Implementing the Configuration for Claude Code:

Because Claude Code operates in a terminal environment and its specific MCP configuration method might depend on its version or how you've set it up, you should consult the official Claude Code documentation or experiment with common MCP configuration file locations for agents, such as a .mcp.json or similar in your home directory, or a dedicated configuration directory.

The most authoritative source for client-side MCP setup details, including how to integrate this JSON block into your Claude Code environment, is the official ToolYour developers guide for MCP: ToolYour MCP Setup Guide. This page provides the latest instructions for various MCP-compatible clients.

Once this configuration is correctly placed and loaded by your Claude Code instance, the agent will be able to discover and use ToolYour's API-backed tools.

Test solve_task from the CLI

After configuring your Claude Code MCP server, the next step is to test its integration by attempting to use a ToolYour tool. The primary agent tool exposed by ToolYour's MCP server is solve_task. This meta-tool allows the agent to articulate a high-level goal, and ToolYour's backend intelligently routes it to the most appropriate specific tool or workflow.

Important Note on solve_task Maturity: solve_task is currently in an early phase. Its routing capabilities, workflow orchestration, and tool coverage are continuously improving as ToolYour upgrades existing tools and integrates new MCP workflows. This means that while it's powerful for experimentation and many common tasks, you might frequently observe a status: suggest response. A suggest status indicates that the agent has identified potential tools or actions but requires further confirmation or input before execution.

For production-critical flows, it is recommended to use the more explicit sequence of discover_tools to find available tools, get_tool_schema to understand their required inputs, and invoke_tool for direct execution. This gives you more control and predictability until solve_task's autonomous routing matures further.

Testing with Claude Code (Conceptual Example):

Given Claude Code's terminal nature, you would typically interact with it by providing prompts or commands. The exact CLI syntax for invoking tools directly via an MCP server with Claude Code isn't standardized across all implementations, but the intent remains the same: communicate a task to the agent that it can delegate to ToolYour.

Conceptually, you might ask Claude Code a question that implies a ToolYour tool usage, or directly instruct it to use a tool if your Claude Code interface allows it.

For example, if you want to summarize an audio file using ToolYour's ai-audio-summarizer tool, your interaction with Claude Code might look like this (this is a placeholder for how Claude Code might expose MCP tools):

# Conceptual Claude Code CLI interaction for solve_task
claude-code agent run --goal "Summarize the provided audio file `speech.mp3` for key points." --context-file speech.mp3

Or, if your Claude Code client provides a more direct way to interact with named MCP servers and their tools, it might be closer to:

# Another conceptual interaction:
claude-code --tool toolyour:solve_task --input '{"goal": "Summarize the key points from the provided meeting notes document. The document is named `meeting_notes.docx`."}' --file meeting_notes.docx

In this example, your Claude Code agent would send the solve_task goal to the ToolYour MCP server. ToolYour's backend would then attempt to route this request to the ai-audio-summarizer or another relevant tool if the context file was, for instance, a .docx for text-stats.

Expected Outcomes:

• Success (status: success): If solve_task successfully identifies and executes a tool, you'll receive the output of that tool. For example, a summary of your audio file.
• Suggestion (status: suggest): It's common for solve_task in its early phase to return suggestions for tools it could use, or ask for more clarification. This is an opportunity for you to refine your prompt or confirm the suggested action.
• Error (status: error): If there's an issue with your API key, network, or an unsupported goal, you'll receive an error.

Remember, running a tool through solve_task (or invoke_tool directly) consumes your ToolYour quota. Browsing available tools or receiving suggestions that don't result in execution do not count against your allowance.

You can explore the full catalog of ToolYour's API-backed tools that are available via MCP at ToolYour AI Tools and ToolYour Digital Tools.

Quota and Troubleshooting

Understanding your ToolYour quota and common troubleshooting steps is essential for a smooth development experience with Claude Code.

Quota Reminders:

• Free Plan Limits: As a reminder, the free plan includes 500 successful requests per month and a rate limit of 5 successful requests per minute. These limits are shared across all API keys under your account and across both REST API and MCP usage.
• Reset Schedule: Your monthly request quota resets on the 1st of each calendar month.
• Quota Exhaustion: If you exceed your monthly request limit, ToolYour will block further requests until your plan is upgraded or the next calendar month begins. There is no "pay-as-you-go" or overage billing on ToolYour plans; requests are simply capped.
• What Counts: Only successful tool calls through MCP (or REST API) count towards your quota. Actions like browsing the ToolYour tool catalog via discover_tools or receiving suggestions from solve_task that do not lead to tool execution do not consume quota. Invalid requests that fail due to malformed input or authentication issues also generally do not count.

Common Troubleshooting Steps:

1. API Key Validity:

   • Issue: Requests are failing with authentication errors.
   • Solution: Double-check that the ty_YOUR_KEY in your Claude Code MCP configuration exactly matches the API key in your ToolYour Dashboard. Ensure there are no leading or trailing spaces. The header should be X-Api-Key.

2. Network Connectivity:

   • Issue: Requests time out or cannot reach the server.
   • Solution: Verify that your Claude Code environment has internet access and can reach https://api.toolyour.com/mcp. Check for any firewall rules or proxy settings that might be blocking the connection.

3. Quota Limits Hit:

   • Issue: Requests are blocked despite correct configuration and valid API key.
   • Solution: Check your usage in the ToolYour Dashboard. If you've reached your monthly or per-minute limit, you will need to wait for the next calendar month for your quota to reset or consider upgrading your plan if continuous high-volume usage is required.

4. solve_task Behavior:

   • Issue: solve_task frequently returns status: suggest instead of directly executing.
   • Solution: This is expected due to solve_task being in an early phase. Refine your prompt to be more specific, or consider using discover_tools followed by get_tool_schema and invoke_tool for direct, predictable execution, especially for critical tasks.

5. Tool-Specific Errors:

   • Issue: A specific tool call fails even after successful MCP connection.
   • Solution: Review the error message returned by the tool. It might indicate missing required parameters, invalid input formats, or other tool-specific issues. Refer to the specific tool's schema obtained via get_tool_schema for input requirements.

For a comprehensive guide to error codes and common issues, visit ToolYour Developer Docs: Errors.

FAQ

Q: What is the Model Context Protocol (MCP)? A: The Model Context Protocol (MCP) is an open standard that allows AI agents to discover, understand, and use external tools and services, expanding their capabilities beyond their core reasoning. You can learn more at modelcontextprotocol.io.

Q: How does ToolYour's MCP server work with Claude Code? A: ToolYour provides a remote MCP server (https://api.toolyour.com/mcp) that Claude Code can connect to using a specific configuration. Once connected, Claude Code can send high-level goals (via solve_task) or specific tool invocations to ToolYour, which then processes them using its 200+ API-backed tools.

Q: Is solve_task free to use with ToolYour? A: Browsing the ToolYour catalog and receiving tool suggestions from solve_task without execution are free and do not consume quota. However, any successful execution of a tool via solve_task (or invoke_tool) counts towards your monthly ToolYour quota, similar to REST API calls.

Q: What are the limits of the ToolYour free plan for MCP usage? A: The free plan includes 500 successful requests per month and 5 successful requests per minute. This quota is shared across all ToolYour surfaces (REST API, MCP) and all API keys on your account. The quota resets on the 1st of each calendar month.

Q: Why does solve_task often return status: suggest? A: solve_task is currently in an early phase. While it's continuously improving, it may frequently return status: suggest when it identifies potential actions but requires further clarification or confirmation before proceeding with execution. For critical workflows, it's recommended to use discover_tools, get_tool_schema, and invoke_tool for more precise control.

Q: Can I use all ToolYour tools via MCP? A: ToolYour's MCP server exposes API-backed tools. Some browser-only tools that do not have an underlying API interface are not available via MCP. ToolYour offers 200+ tools through its REST API and MCP.

Q: Where can I find my ToolYour API key? A: You can generate and manage your API keys from your ToolYour dashboard: ToolYour API Keys. Your key will start with the ty_ prefix.

Conclusion

Integrating ToolYour's MCP server with Claude Code unlocks a vast array of utility tools, empowering your agent to perform complex tasks directly from your terminal environment. By following the steps outlined – from securing your API key and configuring the MCP endpoint to testing solve_task – you can significantly enhance Claude Code's capabilities. Remember to monitor your quota and consult the ToolYour developer documentation for any advanced needs.

Start exploring the possibilities today:

• Get your API key: ToolYour Dashboard / API Keys
• Learn more about MCP integration: ToolYour Developers / MCP
• Browse available tools: ToolYour AI Tools and ToolYour Digital Tools