Skip to content
Gemini CLI

Shared memory for your team’s Gemini CLI

MemScope is a shared memory your team’s Gemini CLI sessions read and write over a remote MCP server. Each person’s Gemini CLI recalls only the memories that person is permitted to see.

The problem

Gemini CLI starts each session with the context files it finds and whatever you type. Something one teammate learned yesterday reaches another only if someone writes it into a file.

Files travel with the repository, so everyone who clones it reads them all. Context meant for one team has to stay out of them or be shown to everyone.

How MemScope fits

Gemini CLI connects to MemScope with an httpUrl entry in its settings and /mcp auth memscope. Sessions can save findings and search the team’s memories, and each search returns what that person may read. AI suggests labels such as sensitivity and topic, but a deterministic policy decides access. Organization admins get no read bypass, MemScope staff see XXX unless the customer grants time-limited, audited support access, and every change is audited.

How access is decided and audited

Install in Gemini CLI

Add the server to Gemini CLI settings.

~/.gemini/settings.json

{
  "mcpServers": {
    "memscope": {
      "httpUrl": "https://mcp.memscope.ai/mcp"
    }
  }
}

In Gemini CLI run /mcp auth memscope.

New to MemScope? Create an account at https://memscope.ai/signup, choose your organization and approve. The same steps, for every client, are on the MCP setup page.

What GEMINI.md files do

Gemini CLI reads GEMINI.md context files: a global ~/.gemini/GEMINI.md, files in your workspace directories and their parents, and component-level files loaded when a tool touches that directory. It concatenates them and sends them with every prompt; /memory show displays the result and /memory reload rescans. You write the files. Those docs do not say whether they are shared across a team and describe no per-person access.

Source: Gemini CLI docs: GEMINI.md context files, checked November 2026.

Compare: How MemScope compares with built-in memory and repo files

Example: same question, two answers

An illustration. Two memories exist: one shared with the whole organization, one limited to the Infra team.

  • Shared with everyone

    Set CACHE_MODE to off when debugging locally.

  • Infra team only

    Production CACHE_MODE changes go through the Infra on-call.

Two people ask Gemini CLI: “How do I turn the cache off?”

  • A teammate on Infra

    Gets the local setting and the production process.

  • A teammate on another team

    Gets the local setting. The Infra-only note is not returned.

Who it is for

Teams whose developers use Gemini CLI in terminals and want its sessions to share what they learn with other agents, each person seeing only what they may.

Price

$15 per connected user per month after a 60-day trial with 3 connected users. You pay only for people who connect an AI client.