MCP Integration

Connect AI to Your A/B Tests

Use the Model Context Protocol (MCP) to manage experiments, analyze results, and automate your testing workflow from Claude Code, Claude Desktop, or ChatGPT. MCP access is currently in private beta.

Get Your API Key REST API Reference

The customer MCP package is in private beta. The public REST API is available today, and the exact MCP install command only appears after Otter A/B enables access for your account.

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources. Instead of copy-pasting API responses, your AI assistant can directly manage your Otter A/B experiments.

Conversational

Ask in plain English: "Create a redirect test on the pricing page" and the AI handles the API calls.

Same API Key

Use the same account-scoped API key you already created for the REST API. No additional auth needed.

Scoped & Safe

API key scopes and project restrictions apply to MCP too. Read-only keys stay read-only.

Prerequisites

Before connecting MCP, you need an Otter A/B API key.

Create an API key

Go to Settings → API & MCP in your Otter A/B dashboard and create a new API key. Select the scopes you want the AI assistant to have access to.

Open Settings

Copy the token

The full token (e.g. oab_live_abc123_sk_...) is only shown once. Store it in a safe place.

Choose recommended scopes

For full MCP functionality, we recommend these scopes:

projects:readprojects:writeexperiments:readexperiments:writeresults:readaccount:read

For read-only access (results inspection only), use: projects:read, experiments:read, results:read

Claude Code

Connect Otter A/B to Claude Code (the CLI) once MCP beta access has been enabled for your account.

Confirm beta access

The public npm package is not published yet. After Otter A/B enables MCP for your account, copy the exact install command from Settings → API & MCP or your beta invite:

# After beta access is enabled:
# copy the exact command from Settings → API & MCP
claude mcp add otterab -- <command provided by Otter A/B>

Set your API key

Set the API key as an environment variable. You can do this globally or per-project:

# Option A: Set in your shell profile (~/.zshrc or ~/.bashrc)
export OTTERAB_API_KEY="oab_live_abc123_sk_9f8e7d..."

# Option B: Add to your project's .claude/settings.json
claude mcp add otterab -e OTTERAB_API_KEY=oab_live_abc123_sk_9f8e7d... -- <beta command from Otter A/B>

Start using it

Open Claude Code and start asking about your experiments:

> claude

You: List my running experiments on the Marketing Site project
Claude: I found 3 running experiments on "Marketing Site":
  1. Pricing Hero Test - running since March 15
  2. Homepage CTA Color - running since March 20
  3. Nav Redesign v2 - running since March 25

You: How is the Pricing Hero Test performing?
Claude: The "Pricing Hero Test" is performing well:
  - Control: 8.36% conversion rate (1,412 visitors)
  - Clearer Headline: 10.72% conversion rate (1,408 visitors)
  - Improvement: +28.2% with 96.1% confidence
  The "Clearer Headline" variant is winning with statistical significance.

Claude Desktop & Cowork

Connect Otter A/B to Claude Desktop or Claude Cowork after Otter A/B enables your private beta access.

Open Claude Desktop settings

Open Claude Desktop and navigate to Settings → Developer → Edit Config. This opens your claude_desktop_config.json file.

Add the beta MCP server configuration

Add the command details from Settings → API & MCP or your beta invite to the config file:

{
  "mcpServers": {
    "otterab": {
      "command": "<provided after beta enablement>",
      "args": ["<provided after beta enablement>"],
      "env": {
        "OTTERAB_API_KEY": "oab_live_abc123_sk_9f8e7d..."
      }
    }
  }
}

Replace the placeholders with the MCP command shown for your account in Settings → API & MCP.

Restart Claude Desktop

Fully quit and reopen Claude Desktop. You should see the Otter A/B tools available in the tools menu (hammer icon).

Start a conversation

Ask Claude to interact with your experiments:

"Show me the results for my pricing page experiment"

"Create a redirect test that sends /pricing traffic to /pricing-v2"

"Pause the Homepage CTA experiment"

ChatGPT

Connect Otter A/B to ChatGPT using the private beta MCP bridge setup or the public REST API.

Install the MCP bridge

ChatGPT supports MCP servers through a local bridge. Install it globally:

npm install -g @anthropic-ai/mcp-chatgpt-bridge

Configure the private beta server

Use the MCP command Otter A/B shares after your account is enabled:

{
  "servers": {
    "otterab": {
      "command": "<provided after beta enablement>",
      "args": ["<provided after beta enablement>"],
      "env": {
        "OTTERAB_API_KEY": "oab_live_abc123_sk_9f8e7d..."
      }
    }
  }
}

Alternative: Direct REST API

You can also use the Otter A/B REST API directly with ChatGPT Actions or custom GPTs. Point the action schema at the API base URL and provide your bearer token:

# ChatGPT Action configuration
openapi: "3.1.0"
info:
  title: Otter A/B API
  version: v1
servers:
  - url: https://www.otterab.com/api/v1
paths:
  /projects:
    get:
      operationId: listProjects
      summary: List all projects
      security:
        - bearerAuth: []
  /projects/{project_id}/experiments:
    get:
      operationId: listExperiments
      summary: List experiments for a project
      parameters:
        - name: project_id
          in: path
          required: true
          schema:
            type: string
      security:
        - bearerAuth: []
components:
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer

Available Tools

Private beta MCP access exposes 17 tools organized by category. Each tool maps to a customer job, not a raw CRUD operation.

Projects

list_projectsprojects

List all projects accessible to this API key.

get_projectprojects

Get details for a specific project by ID.

create_projectprojects

Create a new project with a name, URL, and platform.

Experiments

list_experimentsexperiments

List experiments for a project, optionally filtered by status.

get_experimentexperiments

Get full experiment details including variants, goals, and targeting.

create_experiment_draftexperiments

Create a new experiment with variants, goals, and targeting in one call.

update_experiment_draftexperiments

Update a draft experiment. Replaces variants and goals.

start_experimentexperiments

Start a draft experiment. Requires 2+ variants and 1+ goals.

pause_experimentexperiments

Pause a running experiment.

resume_experimentexperiments

Resume a paused experiment.

complete_experimentexperiments

Complete a running or paused experiment.

archive_experimentexperiments

Archive a completed experiment.

quick_create_redirect_experimentexperiments

One-step creation of a redirect experiment with source URL, redirect URL, and goal.

quick_create_visual_experimentexperiments

One-step creation of a visual experiment with element changes and goal.

Results

get_experiment_resultsresults

Get results for an experiment across all goals with optional segment filters.

analyze_experimentresults

Get an AI-friendly summary of experiment results for the primary goal.

Account

list_targeting_fieldsaccount

List available targeting fields and operators for building rules.

Usage Examples

Here are common workflows you can accomplish with the MCP integration.

Create a redirect experiment

Ask your AI assistant in natural language:

"Create a redirect experiment on the pricing page that sends half of traffic to /pricing-v2 and measures signup completion."

The AI calls:

{
  "tool": "quick_create_redirect_experiment",
  "input": {
    "project_id": "42",
    "name": "Pricing V2 Redirect Test",
    "source_url": "https://example.com/pricing",
    "redirect_url": "https://example.com/pricing-v2",
    "goal": {
      "goal_type": "pageview",
      "name": "Signup Complete",
      "config": { "url": "https://example.com/signup/success" }
    }
  }
}

Analyze experiment results

Ask for a results breakdown:

"How is the Homepage CTA experiment performing? Show me the results by country."

The AI calls:

{
  "tool": "get_experiment_results",
  "input": {
    "project_id": "42",
    "experiment_id": "92",
    "segment": { "country": "US" }
  }
}

Manage experiment lifecycle

Control experiments with simple commands:

"Pause the pricing test"

"Resume the nav redesign experiment"

"Complete the homepage CTA test and show me the final results"

The AI calls the appropriate lifecycle tool:

{
  "tool": "complete_experiment",
  "input": {
    "project_id": "42",
    "experiment_id": "92"
  }
}

Troubleshooting

Common issues and how to resolve them.

"Tools not showing up in Claude Desktop"

Make sure you fully quit and reopened Claude Desktop after editing the config. Check that the config file is valid JSON and that you copied the beta command exactly as shown in Settings → API & MCP.

"Authentication failed" or "invalid_api_key"

Double-check your API key is correct and not revoked. Make sure the OTTERAB_API_KEY environment variable is set. Keys starting with oab_test_ only work with test data.

"Missing required scope"

Your API key doesn't have the required scopes for that action. Create a new key with the needed scopes in Settings > API & MCP.

"Rate limit exceeded"

You've hit your plan's API rate limit. Wait a minute and try again, or upgrade your plan for higher limits. Starter: 100/min, Growth: 500/min, Scale: 2,000/min.

"npx command not found"

Make sure Node.js (v18+) is installed. The beta MCP command may rely on npm/npx, depending on the package Otter A/B enabled for your account.