Gemini

Connect Crow to Google's Gemini — both the Gemini CLI and Gemini in Google AI Studio/Enterprise.

Prerequisites

• Crow gateway deployed and healthy (Cloud Deploy Guide)

Gemini CLI

The Gemini CLI supports remote MCP servers via configuration.

Setup Steps

1. Edit ~/.gemini/settings.json:

   {
     "mcpServers": {
       "crow-memory": {
         "url": "https://your-gateway.onrender.com/memory/mcp"
       },
       "crow-research": {
         "url": "https://your-gateway.onrender.com/research/mcp"
       },
       "crow-tools": {
         "url": "https://your-gateway.onrender.com/tools/mcp"
       }
     }
   }

2. Start Gemini CLI — it will auto-discover the OAuth metadata and prompt for authorization.

Google AI Studio

Google AI Studio supports MCP servers for tool use in chat. Configuration is available through the interface when creating a new chat or agent.

Setup Steps

1. Open Google AI Studio
2. Create a new chat or agent
3. In the tools section, add an MCP server
4. Enter the Streamable HTTP URL:

   https://your-gateway.onrender.com/memory/mcp

5. Complete the OAuth authorization flow

Transport

• Type: Streamable HTTP
• Protocol: 2025-03-26
• Auth: OAuth 2.1 (automatic discovery)

Cross-Platform Context

Load Crow's shared behavioral context on Gemini:

"Use the crow_get_context tool with platform set to gemini"

Or fetch via HTTP: GET https://your-gateway.onrender.com/crow.md?platform=gemini

Memories and research stored from any platform are shared. See the Cross-Platform Guide.

Verification

Ask Gemini:

"Store a memory that Gemini is connected to Crow."

Then verify:

"Search memories for 'Gemini'."