Setup guideAbout 1 minuteLast checked 12 October 2026

Connect CaptableAI to VS Code

GitHub Copilot in agent mode. Once it's connected, ask about your equity in plain language, upload documents, model a round and export the result, all from the same chat.

Short answer

Click Install in VS Code below, or run code --add-mcp with the CaptableAI URL. Start the server, sign in when prompted, and pick CaptableAI from the tools list in Copilot's Agent mode.

Install in VS Code

What you need

✓VS Code with GitHub Copilot. MCP tools are used from agent mode.
✓A CaptableAI company where you are an owner or admin.

Set it up

Install with one click

The button above opens VS Code and adds the server to your user configuration.

Or from the terminal

This adds the same entry without opening any settings.

Terminal

code --add-mcp '{"name":"captable-ai","type":"http","url":"https://mcp.captable.ai/mcp"}'

Or per workspace

To share it with everyone working in a repository, add it to .vscode/mcp.json.

.vscode/mcp.json

{
  "servers": {
    "captable-ai": {
      "type": "http",
      "url": "https://mcp.captable.ai/mcp"
    }
  }
}

Start, trust and sign in

Run MCP: List Servers from the command palette, select captable-ai and choose Start. Confirm you trust the server, then sign in when VS Code opens your browser.

Use it in Agent mode

Open Chat, switch to Agent and tick CaptableAI in the tools picker.

Check it worked

MCP: List Servers should show captable-ai as running. In Agent mode, ask “Which CaptableAI tools can you use?”

First things to ask

“Export the current cap table to Excel and save it in ./reports.”

“Write a script that checks our cap table totals against last month's export.”

“Summarise every equity event since January.”

Troubleshooting

Do I need GitHub Copilot?

Yes. VS Code uses MCP tools from Copilot's agent mode.

The server won't start in my company's VS Code.

Organisations can disable MCP through policy. Ask whoever manages your Copilot settings to allow it.

User or workspace configuration?

User configuration follows you across projects. .vscode/mcp.json is better when a whole team works in the same repository; it holds no secrets.

Then: upload your documents

CaptableAI builds your cap table by reading your legal documents. These seven habits keep it fast and accurate. Share them with VS Code.

01 Upload everything first +

Add all of your cap table documents up front, in one batch where possible: incorporation documents, SAFEs, stock and option agreements, board consents. PDF, DOCX, XLSX, DOC and images are supported.

02 Wait for classification +

Each document is read and classified, so the system knows its type and effective date. Don't trigger extraction until every document shows CLASSIFIED.

03 Check the classification dates +

The effective date decides processing order. If one looks wrong, fix it before extraction so the cap table is built in the right sequence.

04 Trigger extraction once +

Extraction doesn't start automatically for agent uploads. Trigger it a single time after everything is classified; the whole queue is picked up.

05 Let the queue run +

Documents are extracted one at a time, earliest first, so later events build on earlier ones. Expect a few minutes per document and don't re-trigger.

06 Validation runs on its own +

When the last document finishes, a company-wide validator checks every extracted object and repairs cross-document issues. There's nothing to trigger.

07 Review, then commit +

When every document reaches PROCESSED, review the drafts and commit. The first commit activates your cap table.