Claude Code

Add your institute’s server to Claude Code with one command, then sign in from the /mcp menu.

Claude Code is Anthropic’s coding agent for the terminal and IDEs. It connects to Vacademy MCP over HTTP and signs in through your browser.

Commands may change. These steps match Anthropic’s Claude Code MCP documentation as of October 2026. See Connect Claude Code to tools via MCP.

Add the server

Copy the server URL from Settings → MCP Server in your admin portal, then run:

claude mcp add --transport http vacademy https://<your backend>/ai-service/mcp/i/<institute_id>

vacademy is just the name Claude Code shows for this server. You can pick another, such as your institute’s short name.

Choose where it is saved

Add --scope to choose who gets the server:

Scope	Flag	Who gets it	Saved in
Local (default)	--scope local	You, in this project only	~/.claude.json
User	--scope user	You, in every project	~/.claude.json
Project	--scope project	Everyone who uses this repository	.mcp.json in the project root

For example, to have it in every project:

claude mcp add --transport http --scope user vacademy https://<your backend>/ai-service/mcp/i/<institute_id>

A project-scoped .mcp.json looks like this. It holds only the URL, never a token, so it is safe to commit. Each teammate still signs in as themselves.

.mcp.json

{
  "mcpServers": {
    "vacademy": {
      "type": "http",
      "url": "https://<your backend>/ai-service/mcp/i/<institute_id>"
    }
  }
}

Sign in

1. Open the MCP menu

   Start Claude Code and type /mcp. Pick the vacademy server and choose to authenticate.

   From the shell, claude mcp login vacademy does the same.

2. Approve in your browser

   A browser tab opens on your institute’s admin portal. Sign in with your own account and click Approve.

3. Check the connection

   Back in Claude Code, ask:

   Who am I connected as?
   

   Claude Code calls the whoami tool to answer. In /mcp, the vacademy server should show as connected, with its tools listed. Only whoami? Every area is Off for your role. Ask an admin to set one to View.

No client ID is needed. Claude Code registers itself and listens for the sign-in on your own machine (localhost), which the server accepts.

Manage it

claude mcp list             # servers and their status
claude mcp get vacademy     # details for this server
claude mcp logout vacademy  # sign out (clear stored tokens)
claude mcp remove vacademy  # remove the server and its stored tokens

Inside Claude Code, /mcp shows the status and lets you clear authentication.

Good uses in Claude Code

Claude Code is handy when you work with files on your machine. For example:

Read ./brochure.pdf and build a draft landing page for our NEET 2027 batch from it. Use our brand colours from whoami. Give me the editor link when it's done.

Here is a Figma link for our new homepage: <figma link>. Follow your Figma playbook and save it as a draft site.

Turn ./syllabus.md into a draft course with one chapter per heading, then run the course review and fix what it reports.

These need the matching Edit levels. Website and course edits are saved as drafts that an admin publishes.

Prompts and playbooks

The server also offers a ready-made prompt and design playbooks. They appear only when the Website area is on for your role.

• Prompt as a slash command. If you named the server vacademy:

  /mcp__vacademy__figma_to_site https://www.figma.com/design/<file>/...
  

• Playbooks as resources. Type @ and pick one, for example @vacademy:vacademy://playbook/figma.

Claude Code reads the Figma file with its own Figma connection, so connect your Figma tools in Claude Code too. See Prompting tips.

Using a fixed client ID

You only need this if your organisation requires every app to use a pre-registered client.

1. An admin creates a client ID under Settings → MCP Server → Client IDs for other apps, with the redirect URL http://localhost:<port>/callback. Pick a free port, for example 8080.
2. Add the server with that client ID and port. There is no client secret.

claude mcp add --transport http \
  --client-id <client id> --callback-port 8080 \
  vacademy https://<your backend>/ai-service/mcp/i/<institute_id>