fix: Set Gemini 2.5 Pro Preview as default and improve portability

Changes: - Restored Gemini 2.5 Pro Preview as the default model - Removed hardcoded paths from claude_config_example.json - Added MCP_DISCOVERY.md explaining how Claude discovers MCP servers - Updated README with natural language usage examples The server now defaults to the most capable Gemini 2.5 Pro Preview model as requested, and all paths are now relative for better portability. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
2025-06-08 19:49:08 +04:00
parent 54d9bb1ee7
commit b86c42cf3a
4 changed files with 106 additions and 5 deletions
--- a/MCP_DISCOVERY.md
+++ b/MCP_DISCOVERY.md
@@ -0,0 +1,93 @@
+# How Claude Discovers and Uses MCP Servers
+
+## MCP Server Discovery
+
+When you configure an MCP server in Claude Desktop, Claude automatically discovers its capabilities through the MCP protocol:
+
+1. **On Startup**: Claude connects to all configured MCP servers
+2. **Tool Discovery**: Claude calls `list_tools()` to discover available tools
+3. **Schema Understanding**: Each tool provides its input schema, which Claude uses to understand how to call it
+
+## How This Gemini Server Appears in Claude
+
+Once configured, this Gemini MCP server provides three tools that Claude can use:
+
+### 1. `gemini_chat`
+- Claude sees this as a way to chat with Gemini
+- You can invoke it naturally: "Ask Gemini about...", "Use Gemini to..."
+
+### 2. `gemini_analyze_code`
+- Claude recognizes this for code analysis tasks
+- Triggered by: "Use Gemini to analyze this file", "Have Gemini review this code"
+
+### 3. `gemini_list_models`
+- Lists available models
+- Usually called automatically when needed
+
+## Natural Language Usage
+
+Claude is smart about understanding your intent. You don't need special syntax:
+
+### Examples that work:
+- "Ask Gemini what it thinks about quantum computing"
+- "Use Gemini to analyze the file /path/to/large/file.py"
+- "Have Gemini review this code for security issues"
+- "Get Gemini's opinion on this architecture"
+- "Pass this to Gemini for extended analysis"
+
+### What happens behind the scenes:
+1. Claude recognizes keywords like "Gemini", "analyze", "review"
+2. Claude determines which tool to use based on context
+3. Claude extracts parameters (files, questions, etc.) from your request
+4. Claude calls the appropriate MCP tool
+5. Claude presents the response back to you
+
+## Configuration in Claude Desktop
+
+### macOS
+Add to `~/Library/Application Support/Claude/claude_desktop_config.json`:
+
+```json
+{
+  "mcpServers": {
+    "gemini": {
+      "command": "/path/to/gemini-mcp-server/venv/bin/python",
+      "args": ["/path/to/gemini-mcp-server/gemini_server.py"],
+      "env": {
+        "GEMINI_API_KEY": "your-api-key-here"
+      }
+    }
+  }
+}
+```
+
+### Windows
+Add to `%APPDATA%\Claude\claude_desktop_config.json`
+
+### After Configuration
+1. Restart Claude Desktop
+2. Claude will automatically connect to the Gemini server
+3. You'll see "gemini" in the MCP servers list (bottom of Claude interface)
+4. Start using natural language to invoke Gemini!
+
+## Verification
+
+To verify the server is connected:
+1. Look for the MCP icon in Claude's interface
+2. Ask Claude: "What MCP tools are available?"
+3. Claude should list the Gemini tools
+
+## Troubleshooting
+
+If Claude doesn't recognize Gemini commands:
+1. Check the MCP server icon shows "gemini" as connected
+2. Verify your API key is set correctly
+3. Check Claude's logs for connection errors
+4. Try restarting Claude Desktop
+
+## Integration with Claude Code
+
+In Claude Code, the integration is even more seamless:
+- Large file handling is automatic
+- Claude will suggest using Gemini when hitting token limits
+- File paths are resolved relative to your workspace
--- a/README.md
+++ b/README.md
@@ -2,6 +2,15 @@

 A Model Context Protocol (MCP) server that enables integration with Google's Gemini models, optimized for Gemini 2.5 Pro Preview with 1M token context window.

+## How It Works with Claude
+
+Once configured, Claude automatically discovers this server's capabilities. You can use natural language to invoke Gemini:
+- "Ask Gemini about..."
+- "Use Gemini to analyze this file..."
+- "Have Gemini review this code..."
+
+See [MCP_DISCOVERY.md](MCP_DISCOVERY.md) for detailed information about how Claude discovers and uses MCP servers.
+
 ## Features

 - **Chat with Gemini**: Send prompts to Gemini 2.5 Pro Preview by default
@@ -115,7 +124,7 @@ When working with large files in Claude Code, you can use the Gemini server like

 ## Models

-The server defaults to `gemini-2.5-pro-preview-06-05` which supports:
+The server defaults to `gemini-2.5-pro-preview-06-05` (the latest and most capable model) which supports:
 - 1 million token context window
 - Advanced reasoning capabilities
 - Code understanding and analysis
--- a/claude_config_example.json
+++ b/claude_config_example.json
@@ -1,8 +1,8 @@
 {
  "mcpServers": {
    "gemini": {
-      "command": "/Users/fahad/Developer/gemini-mcp-server/venv/bin/python",
-      "args": ["/Users/fahad/Developer/gemini-mcp-server/gemini_server.py"],
+      "command": "/path/to/gemini-mcp-server/venv/bin/python",
+      "args": ["/path/to/gemini-mcp-server/gemini_server.py"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key-here"
      }
--- a/gemini_server.py
+++ b/gemini_server.py
@@ -18,8 +18,7 @@ import google.generativeai as genai


 # Default to Gemini 2.5 Pro Preview with maximum context
-# Note: 2.5 Pro Preview has restrictions, falling back to 1.5 Pro for better reliability
-DEFAULT_MODEL = "gemini-1.5-pro-latest"  # More reliable, still has large context
+DEFAULT_MODEL = "gemini-2.5-pro-preview-06-05"
 MAX_CONTEXT_TOKENS = 1000000  # 1M tokens