Skip to content

MCP Builder

Enable Agents to design and create custom MCP tool servers through guided development using the FastMCP framework.

Core Capabilities

MCP Builder provides complete MCP server development through AI collaboration:

Capability Benefit Example
Requirements Gathering Agent asks clarifying questions to understand tool needs User says "weather API tool" → Agent asks about data sources, API keys, features
Development Guidance Provides complete FastMCP specification with examples and best practices Learn how to use FastMCP decorators, define tools, handle errors
Configuration as Files Tool designs are Markdown files with structured XML tags mcp/weather_api.md created with <folder_name>, <script_file>, <dependencies> tags
Automatic Setup Parses design and creates tool folder, installs dependencies, configures virtual environment From document → working MCP server ready to use
Iterative Refinement Modify designs based on feedback until satisfied Draft → User feedback → Adjustment → Repeat until perfect

Why Use MCP Builder

  • Natural language conversation to design tools
  • Agent asks questions to understand requirements
  • Provides FastMCP specification with templates
  • Automatic tool creation, dependency installation, registration
  • Iterative improvement based on feedback

Collaboration Pattern:

  1. User states need in natural language
  2. Agent asks clarifying questions (API sources, features, requirements)
  3. Gets FastMCP specification to understand framework
  4. Designs tool code and configuration
  5. Creates Markdown file with structured XML tags
  6. User provides feedback, Agent adjusts design
  7. Repeat until user approves
  8. Registers tool - automatic setup and configuration

Available Operations

Operation Purpose When to Use
get_mcp_spec Get complete FastMCP development specification Starting new MCP tool design to understand framework
register_mcp Parse tool document and register to system After user approves final design

Note: MCP Builder works with File Tool to create/edit tool configuration documents.

Common Workflows

Creating New MCP Tool

Stage Action Details
1. Requirements Agent asks questions "What data/services to access?", "Need API keys?", "Input/output format?"
2. Specification Get development guide get_mcp_spec retrieves FastMCP examples, templates, XML tag explanations
3. Design Draft tool code Agent designs tool functions, parameters, error handling using FastMCP
4. Document Create structured file create_file("mcp/weather.md", ...) with <folder_name>, <script_file>, <dependencies>, <readme> tags
5. Review User examines design Agent explains implementation, user provides feedback
6. Refine Iterate on feedback User: "Add caching" → Agent updates with update_file
7. Register Setup and activate register_mcp("mcp/weather.md") → Creates folder, installs packages, registers server

Optimizing Existing Tool

Stage Action Details
1. Export Get current config User or Agent reads existing tool document from mcp/ directory
2. Review User identifies changes "Add retry logic", "Support more data formats"
3. Modify Update code Agent edits Markdown file using update_file or rewrite_file
4. Refine Iterate until satisfied Multiple rounds of user feedback and adjustments
5. Update Re-register register_mcp("mcp/tool_name.md") → Updates tool with new version

Human-AI Collaboration

Agent Responsibilities

  • Ask clarifying questions before designing
  • Explain FastMCP concepts clearly
  • Present draft for review before registering
  • Listen to feedback and adjust accordingly
  • Ensure user satisfaction before finalizing

Integration with Other Tools

+ File Tool: MCP tool configs are Markdown files managed with file operations - create_file to create initial design - read_file to review current config - update_file for targeted changes (small tweaks) - rewrite_file for major revisions (restructuring)

+ System Operations: After creation, tools are available system-wide - list_all_mcps to see registered tools - get_mcp_details to view tool information

Best Use Cases

Good For: - Connecting to external APIs (weather, news, databases) - Custom data processing tools - Domain-specific integrations - Extending Agent capabilities with specialized functions