Connect ContextDrop to Gemini

Before you start

What you need

Custom MCP availability can depend on your Google account, region, and Gemini feature rollout.

Step 1

Open Gemini Connectors

Open Gemini and select Customise from the sidebar. Open the Connectors tab, then click the Custom button beside the search field.

Step 2

Enter the MCP server URL

In the Connect to an MCP server dialog, enter the following URL.

https://mcp.usecontextdrop.com/mcp

Expand Additional settings if needed. In our tested setup, we left the optional Client ID and Client secret fields empty and continued with Next. If Gemini requires credentials in your environment,don't enter unrelated credentials; investigate the OAuth registration requirements first.

Step 3

Authorize Gemini

Review the custom Connected App confirmation and continue to the ContextDrop authorization page.

Step 4

Save the custom app

After authorization, Gemini displays a Save your custom app dialog. Review the discovered tools and click Connect.

Step 5

Configure your active project

In ContextDrop, open Settings → MCP & Integrations and select the project you want Gemini to use as your active MCP workspace.

Gemini's requests will use the authenticated user's active ContextDrop workspace. Ensure the intended project is selected before testing retrieval or saving memories.

Step 6

Configure Gemini instructions

To encourage consistent retrieval and autonomous memory maintenance, add the following instructions to the instruction field you use for your Gemini conversations. The exact settings location may vary by Gemini interface.

Use ContextDrop as the persistent memory layer for project-specific work.

### Retrieval
- At the start of project-specific tasks, retrieve relevant project context and search existing memories.
- Treat memories as supporting context and check for conflicts with the current conversation.
- If nothing relevant is found, continue normally.

### Autonomous memory
- Identify durable project knowledge such as architecture, decisions, conventions, requirements, constraints, and technical discoveries.
- Search existing memories before saving to avoid duplicates.
- Save new durable knowledge using save_memory.
- Update existing memories using update_memory when established knowledge materially changes.
- Ignore temporary debugging details, speculation, secrets, generic programming knowledge, and short-lived tasks.
- Never claim a save or update succeeded unless the tool confirms it.

### Conversational continuity
- Keep memory operations unobtrusive and do not interrupt the user's workflow.
- Continue the original task immediately after a successful memory operation.
- Do not announce routine searches or successful saves.
- Do not explain memory management unless asked.
- Mention memory operations only when clarification is necessary or a meaningful conflict or failure requires attention.

### Security
- Use only the authenticated user's authorized ContextDrop workspace.
- Never access another user's or workspace's memories.

These instructions guide Gemini'sbehavior; they do not guarantee that it will invoke MCP tools on every relevant message. Verify actual tool calls when testing autonomous memory behavior.

Step 7

Verify shared memory retrieval

Start a new Gemini conversation and ask it to retrieve a memory already stored in ContextDrop. For a reliable test, explicitly request a search before testing whether Gemini selects the tool automatically.

Use the ContextDrop MCP app to search for the API validation and authorization convention.

Return the actual matching memory titles and contents. Do not answer from general knowledge, and do not save or update anything.