Documentation for using FastMCP as an MCP Client
Enhancement
We should advertise FastMCP Client as a great way to add MCP tools to your existing code, chat application or AI agent framework.
We should include examples use cases in the docs for this
Claude finished @strawgate's task ββ View job
Investigation Complete
Issue Type: Documentation Enhancement
Recommendation
Create a dedicated "Use Cases" documentation page highlighting FastMCP Client as a powerful way to integrate MCP tools into existing applications. The existing client documentation is technically comprehensive but doesn't effectively advertise the client's value proposition or provide compelling use case examples.
The documentation should emphasize three primary integration patterns:
- AI SDK Integration - Adding MCP tools to AI applications (Anthropic, OpenAI, Gemini)
- Testing & Development - Using the client to test MCP servers during development
- Agent Framework Integration - Building custom AI agents that leverage MCP tools
Findings
Current State
Existing Client Documentation (docs/clients/):
- β Comprehensive technical coverage of Client API
- β Good examples of transports, auth, and operations
- β Covers multi-server configuration
- β Missing: Value proposition / "why use this" framing
- β Missing: Real-world integration examples
- β Missing: Use case-driven examples
Existing Integration Examples:
-
AI SDKs: Anthropic (
docs/integrations/anthropic.mdx), OpenAI (docs/integrations/openai.mdx), Gemini (docs/integrations/gemini.mdx)- These show how to connect FastMCP servers TO the AI SDKs
- β Missing reverse: Using FastMCP Client to ADD tools to AI applications
-
Testing Pattern: Well-documented in AGENTS.md and test files
- Shows in-memory transport pattern
- β Good foundation but buried in developer docs
Gap Analysis:
- The client is positioned as a "programmatic interface" but not as a solution for common problems
- Integration examples show "MCP server β AI SDK" but not "FastMCP Client β your code"
- No dedicated page for "adding MCP to your existing application"
- Use cases are scattered across different docs rather than centralized
Related Items Context
Issue #1839 - "MCP Client from multiple in memory servers"
- Shows user confusion about client capabilities
- Demonstrates need for clearer multi-server examples
- Resolved with proxy pattern explanation
Issue #1878 - "Add more Transformation Examples"
- Shows related need for client usage patterns
- Focused on tool transformation (exposing client methods as tools)
- Complementary to this request (that's about consuming tools, this is about providing them)
PR #2002 - "Additional Tool Transformation Examples"
- Recently added examples for client wrapping patterns
- Shows pattern of improving practical examples
Detailed Action Plan
1. Create New Documentation Page: docs/clients/use-cases.mdx
Structure:
---
title: FastMCP Client Use Cases
description: Real-world examples of integrating MCP tools into your applications
---
## Why Use FastMCP Client?
FastMCP Client provides a programmatic way to add MCP tools to any Python application...
## Use Case 1: Integrating with AI SDKs
### Anthropic SDK
[Example: Adding weather tools to Claude API calls]
### OpenAI SDK
[Example: Adding database tools to GPT applications]
### LangChain
[Example: Creating LangChain tools from MCP servers]
## Use Case 2: Building Custom AI Agents
[Example: Agent that uses multiple MCP servers]
[Example: Agentic loop with tool selection]
## Use Case 3: Testing MCP Servers
[Example: In-memory testing pattern]
[Example: Integration test setup]
## Use Case 4: Chat Applications
[Example: Adding tools to custom chat interface]
[Example: Multi-turn conversations with tool use]
2. Update docs/clients/client.mdx
Add section near the top (after "Creating a Client"):
## Why Use the FastMCP Client?
The FastMCP Client is ideal for:
- **π€ AI Integration**: Add MCP tools to applications built with Anthropic, OpenAI, or other AI SDKs
- **π§ͺ Testing**: Test your MCP servers during development with in-memory connections
- **π€ Agent Frameworks**: Build custom AI agents that leverage multiple MCP servers
- **π¬ Chat Apps**: Enhance chat applications with dynamically loaded tools
For detailed examples of each use case, see [Use Cases](/clients/use-cases).
3. Create Example Code Files
examples/client_integrations/anthropic_client.py:
"""Example: Using FastMCP Client to add tools to Anthropic SDK."""
examples/client_integrations/openai_client.py:
"""Example: Using FastMCP Client to add tools to OpenAI SDK."""
examples/client_integrations/custom_agent.py:
"""Example: Building a custom agent with FastMCP Client."""
4. Update Navigation
In docs/docs.json, add to Clients section:
{
"group": "Essentials",
"icon": "cube",
"pages": [
"clients/client",
"clients/use-cases", // NEW
"clients/transports"
]
}
5. Update README.md
Add to "MCP Clients" section (line ~240):
The FastMCP Client is not just for testing - it's a powerful way to integrate MCP tools into any Python application, including:
- AI applications built with Anthropic, OpenAI, or Gemini SDKs
- Custom AI agent frameworks
- Chat applications requiring dynamic tool loading
Related Files
Documentation Files to Create/Modify
| File | Action | Purpose |
|---|---|---|
docs/clients/use-cases.mdx |
CREATE | New dedicated use cases page |
docs/clients/client.mdx |
MODIFY | Add "Why Use" section with value prop |
examples/client_integrations/anthropic_client.py |
CREATE | Example: MCP tools + Anthropic SDK |
examples/client_integrations/openai_client.py |
CREATE | Example: MCP tools + OpenAI SDK |
examples/client_integrations/custom_agent.py |
CREATE | Example: Custom agent with multiple servers |
docs/docs.json |
MODIFY | Add use-cases to navigation |
README.md |
MODIFY | Enhance MCP Clients section |
Key Reference Files
| File | Relevance | Key Sections |
|---|---|---|
| docs/clients/client.mdx | Current client overview | Lines 1-291: Complete client documentation |
| docs/integrations/anthropic.mdx | Shows serverβSDK pattern | Lines 1-229: Could be adapted for reverse (ClientβSDK) |
| docs/integrations/gemini.mdx | Shows direct client usage | Lines 58-82: Good example of Client + AI SDK |
| README.md | Main project overview | Lines 240-299: MCP Clients section |