Flet MCP Server

Model Context Protocol that serves official Flet resources to AI agents.

Flet MCP Server dynamically fetches and serves official Flet documentation, controls, packages, and ecosystem resources for AI agents and MCP-compatible clients.

Features

• GitHub Tree Sync: Maps documentation in real-time from the Flet repository.
• Intelligent Caching: Uses diskcache for fast responses with 24-hour TTL.
• Smart Search: Fuzzy matching, keyword aliases, and direct path search over documentation.
• Ecosystem Discovery: Finds and verifies official and community Flet packages via PyPI metadata.
• AI-Optimized: Tool definitions designed for LLM understanding with clear usage guidance.
• Configurable: Supports custom Flet repo/branch via environment variables.
• Resource Safe: Shared HTTP client with proper lifecycle management and graceful shutdown.

Tools Included

1. list_flet_controls

List all available Flet UI controls.

2. search_flet_docs(query)

Search the documentation index with fuzzy matching and keyword aliases.

3. get_flet_doc(doc_path)

Get raw Markdown for a specific doc file.

4. list_official_packages()

List official Flet extension packages from the monorepo.

5. search_flet_ecosystem(query)

Search for verified community Flet components on GitHub.

6. get_package_details(package_name)

Fetch version, type classification, and installation info from PyPI.

Client Configuration Examples

🌌 VSCode

Add this to your .vscode/mcp.json:

{
  "servers": {
    "flet-mcp-server": {
      "command": "uvx",
      "args": ["flet-mcp-server"]
    }
  }
}

🌌 Antigravity / Cascade

Add this to your mcp_config.json:

{
  "mcpServers": {
    "flet-mcp-server": {
      "command": "uvx",
      "args": ["flet-mcp-server"]
    }
  }
}

🤖 Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "flet-mcp-server": {
      "command": "uvx",
      "args": ["flet-mcp-server"]
    }
  }
}

💻 Cursor / Windsurf

In your IDE's MCP settings, add a new server:

• Name: Flet MCP
• Type: Command
• Command: uvx flet-mcp-server

💻 OpenCode

Add this to your ~/.opencode/opencode.json or project-level .opencode/opencode.json:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "flet-mcp": {
      "type": "local",
      "command": ["uvx", "flet-mcp-server"],
      "enabled": true
    }
  }
}

For authenticated GitHub API access (higher rate limits), set GITHUB_TOKEN in your environment:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "flet-mcp": {
      "type": "local",
      "command": ["uvx", "flet-mcp-server"],
      "enabled": true,
      "env": {
        "GITHUB_TOKEN": "your-github-token"
      }
    }
  }
}

Environment Variables

Development

Directory Structure

flet-mcp-server/
├── .github/
│   └── workflows/
│       ├── ci.yml
│       └── publish.yml
├── docs/
│   ├── ARCHITECTURE.md
│   └── CONTRIBUTING.md
├── src/
│   └── flet_mcp/
│       ├── services/
│       │   ├── github_docs.py
│       │   ├── packages.py
│       │   └── __init__.py
│       ├── __init__.py
│       ├── exceptions.py
│       ├── http.py
│       ├── main.py
│       └── server.py
├── tests/
│   └── test_fetcher.py
├── LICENSE
├── README.md
├── pyproject.toml
└── uv.lock

Install

git clone https://github.com/Nwokike/flet-mcp-server.git
cd flet-mcp-server
uv sync

Test

uv run pytest

Lint

uv run ruff check .

Changelog

v0.2.0

• Smart Search: Added fuzzy matching, keyword alias index, and multi-strategy search (direct > alias > fuzzy)
• Shared HTTP Client: Single httpx.AsyncClient with proper lifecycle management via server lifespan
• Custom Exceptions: DocNotFoundError, PackageNotFoundError, FetchError, TreeFetchError for better error handling
• Configurable Repo/Branch: FLET_REPO and FLET_BRANCH environment variables for version pinning
• Robust Control Parsing: Set-based deduplication with safer path parsing
• Concurrent Verification: Batched PyPI verification with semaphore-based rate limiting for ecosystem search
• Graceful Shutdown: Server lifespan handler cleans up HTTP connections
• Better Error Handling: All tools catch and format errors gracefully for LLM consumption
• Mocked Tests: Full test suite with mocked HTTP responses for fast, reliable CI
• OpenCode Support: Added OpenCode MCP configuration example to README
• Expanded Fallback: Added flet, flet-cli, flet-desktop, flet-web to fallback package list

• Initial release with 6 MCP tools
• GitHub Tree API integration with diskcache
• PyPI verification for ecosystem packages

License

MIT