Connect Corexi to your IDE.

Open Corexi → Settings → API keys → New API key. Give it a name you will recognise (“Cursor on laptop”), then copy the key. It is shown once.

Keys start with crxi_live_. We store only a salted SHA-256 hash; a lost key is revoked and re-issued.

Replace crxi_live_YOUR_KEY_HERE with the key you just generated.

{
  "mcpServers": {
    "corexi": {
      "url": "https://corexi.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer crxi_live_YOUR_KEY_HERE"
      }
    }
  }
}

Ask your agent to list my Corexi sites. Then open Settings → API keys: the key's Last used column fills in on the first request that reaches Corexi. An IDE that says “connected” while Last used still says Never is not sending the key.

In a GUI client (MCP Inspector, a server manager with an “Edit server” form) the URL alone is not enough: add Authorization: Bearer crxi_live_… in its headers field. Without it the server answers 401 and the client shows it as disconnected.

Once connected, your agent calls these on its own when you ask about UX. You do not invoke them by hand.

• get_ux_rulesNo scan

  Returns Corexi's research-backed UX rules. Filter by category, severity, finding type or free text.

  "What are the rules for designing accessible CTAs?"

• get_ux_checklistNo scan

  Returns a pre-flight checklist for a specific surface (checkout, modal, hero, pricing, navigation…).

  "Give me the checklist before I implement the checkout form."

• review_codeopen beta · token

  LLM review of a snippet against Corexi's 9 categories and the neurodiversity lenses. Pass site_id to inject your design tokens.

  "Review this checkout button JSX before I commit."

• review_screenshotShared quota

  Vision-LLM review of a mockup, Figma export or local screenshot. Validates designs at mockup stage.

  "Review this Figma export of the new pricing page."

• list_sites

  List the sites registered to your Corexi account.

  "What Corexi sites do I have?"

• get_latest_scan

  Fetch the latest scan summary for a site — score, category breakdown, pages, timestamp.

  "What's the UX Score on my pricing page?"

• get_findings

  List prioritized findings for a scan, filterable by severity and status.

  "Show me the top 5 high-severity issues on my homepage."

• trigger_scanscan scope

  Start an Autopilot run on a product you own, from the IDE. It returns a job id straight away. Subject to plan limits and cooldowns.

  "Run Autopilot on my app."

• get_scan_job

  Follow a run started with trigger_scan: its steps as they happen, or wait for the end, without blocking the IDE.

  "Is the run done yet?"

• get_run_diff

  What a run changed against the one before it: fixed, new, back, not seen, still open. Use it after a deploy.

  "What did the last deploy fix?"

Troubleshooting