VSCode + GitHub Copilot

Prerequisites

• VSCode 1.99 or later (MCP support).
• GitHub Copilot or Copilot Chat extension active.

Setup (workspace-scoped)

1. In Draxis, Settings → API Tokens → New token. Name it “VSCode”, pick a scope (read for query-only Copilot Chat, read + write if you want Copilot to push evidence to the AI Drop Zone or trigger integration runs), copy the plaintext.
2. In your VSCode workspace, create .vscode/mcp.json:

   {
     "servers": {
       "draxis": {
         "type": "http",
         "url": "https://app.draxis.ai/api/mcp",
         "headers": {
           "Authorization": "Bearer drx_pat_<tenant>.<token>"
         }
       }
     }
   }

3. Reload the VSCode window. Open the Copilot Chat panel, click the tools icon, and confirm draxis shows in the list.
4. Try a prompt: “@workspace use the draxis tools to list our open risks.”

Setup (user-scoped, OAuth)

For organizations that prefer OAuth-based revocation, configure the server in your User Settings (JSON) instead, and omit the headers. VSCode will surface a sign-in prompt the first time the server is invoked. Pick the tenant in the consent flow and the access tokens are stored in VSCode’s secret store.

Available tools

Same tool surface as every other MCP client. Nine read tools: list_kris, get_kri, list_controls, list_risks, list_outcomes, list_vendors, get_asset_inventory, list_simulations, list_integrations. Two write tools (require read write on a PAT or mcp:write on an OAuth token): submit_dropzone_artifact, run_integration. See the full reference table for what each returns.

Troubleshooting

• Tools picker doesn’t list draxis, run the “MCP: List Servers” command from the palette to see the connection error.
• “Tool call failed: 401”, your PAT was revoked or expired, mint a new one.
• Still stuck? Open a ticket.

Other LLM clients

See Claude Desktop, Claude Code, ChatGPT Connectors, Cursor.