Connect an agent

Give an AI tool access to your GigaRAG Workspace, so it can search, read and write your memos while it works.

ZCode

Z.ai's agentic development environment for GLM models.

Connect with the GigaRAG CLI

ZCode keeps its settings somewhere a command cannot safely rewrite, so the CLI prints exactly what to enter instead. Signing in still happens in your browser, and no key is written into any config file.

npm install -g gigarag
gigarag login
gigarag connect zcode

Paste what it prints, then restart ZCode afterwards. Nothing here expires, and gigarag connect --remove takes GigaRAG back out of everything it touched.

Sign in with GigaRAG

Connect without a key: ZCode opens your browser, you sign in to GigaRAG and pick this workspace.

1.
Add GigaRAG as below, but leave Headers empty
2.
Click Open authorization on the GigaRAG row
3.
Sign in with your GigaRAG account
Your browser opens GigaRAG. Pick the workspace to share and approve. No key is stored in the app, and the connection does not expire.

Or set it up by hand

1.
Open Settings, then MCP Servers, then New MCP Server
Choose HTTP and name it gigarag.
2.
Paste this as the Service URL
Server URL
```
https://mcp.gigarag.com/mcp
```
3.
Under Headers, add Authorization with this value
Authorization
```
Bearer <api-key>
```

Check it worked

Send this once it is connected. An answer drawn from your workspace means the connection is live.

List my GigaRAG buckets.

Written from the official docs: ZCode MCP