# A/B testing MCP servers

> An MCP server lets your own AI agent operate an A/B testing tool. Of the tools we track, 21 document one (fully or partly), plus Duelpilot, whose MCP server is how the whole product is run.

Last reviewed: 2026-10-17 · HTML version: https://duelpilot.com/mcp-ab-testing

## Which A/B testing tools have an MCP server

As documented by each vendor. Each cell links to its source.

| Tool | MCP server | Your agent designs variants | Agent can launch |
|---|---|---|---|
| Duelpilot | Yes — Remote MCP server with 48 tools; OAuth for claude.ai and ChatGPT, bearer token elsewhere. [source](https://duelpilot.com/agents.md) | Yes — Your agent designs 2–4 variants per idea, each with a preview and a device check. [source](https://duelpilot.com/agents.md) | No — A person approves every launch; an owner can switch on an autopilot policy with guardrails, never an agent. [source](https://duelpilot.com/index.md) |
| [Amplitude Experiment](https://duelpilot.com/compare/duelpilot-vs-amplitude-experiment) | Yes — mcp.amplitude.com/mcp. [source](https://amplitude.com/docs/amplitude-ai/amplitude-mcp) | Yes — create\_experiment with custom variants, create\_flags, update\_experiment. [source](https://amplitude.com/docs/amplitude-ai/amplitude-mcp) | Unknown — No dedicated start tool documented; update\_flag covers deployments. [source](https://amplitude.com/docs/amplitude-ai/amplitude-mcp) |
| [Convert](https://duelpilot.com/compare/duelpilot-vs-convert) | Yes — Hosted OAuth connector and the @convertcom/mcp-server package: 16 top-level tools with 113 callable actions when full access is enabled. [source 1](https://www.convert.com/mcp-server/), [source 2](https://support.convert.com/hc/en-us/articles/convert-mcp-server) | Yes — With full access, the agent can create, clone and update experiences and variations; new work is created as a draft. [source](https://www.convert.com/mcp-server/) | Partly — Status changes are possible with full access, but activating or pausing requires explicit confirmation; the local server starts read-only. [source](https://www.convert.com/mcp-server/) |
| [Crazy Egg](https://duelpilot.com/compare/duelpilot-vs-crazy-egg) | Yes — mcp.crazyegg.com/mcp with OAuth. [source](https://support.crazyegg.com/knowledge-base/crazy-egg-model-context-protocol-mcp/) | Unknown — The documented tools are chat and harness tools; A/B-test actions are not described. [source](https://support.crazyegg.com/knowledge-base/crazy-egg-model-context-protocol-mcp/) | Unknown — No test-launch tool documented. [source](https://support.crazyegg.com/knowledge-base/crazy-egg-model-context-protocol-mcp/) |
| [Eppo (Datadog Experiments)](https://duelpilot.com/compare/duelpilot-vs-eppo) | Yes — Datadog MCP Server with the experiments toolset enabled (off by default). [source](https://docs.datadoghq.com/mcp_server/tools/) | Yes — create\_experiment, create\_datadog\_feature\_flag, link\_feature\_flag\_to\_experiment. [source](https://docs.datadoghq.com/mcp_server/tools/) | Yes — start\_experiment, conclude\_experiment and cancel\_experiment tools; no approval gate is documented. [source](https://docs.datadoghq.com/mcp_server/tools/) |
| [Flint](https://duelpilot.com/compare/duelpilot-vs-flint) | Yes — Remote MCP server at mcp.tryflint.com/mcp, added as a Claude connector. [source 1](https://www.flint.com/docs/claude-mcp), [source 2](https://www.flint.com/product/flint-mcp) | Partly — Claude creates and edits pages, not test variants. [source](https://www.flint.com/docs/claude-mcp) | Partly — publish\_site deploys a page to production; there are no experiment tools. [source](https://www.flint.com/docs/claude-mcp) |
| [GrowthBook](https://duelpilot.com/compare/duelpilot-vs-growthbook) | Yes — Hosted at mcp.growthbook.io/mcp or local via @growthbook/mcp. [source](https://docs.growthbook.io/integrations/mcp) | Yes — growthbook\_api\_write can make any write call the REST API allows, including creating experiments. [source](https://docs.growthbook.io/integrations/mcp) | Yes — Optional: the API write tool can start experiments; confirmation can be required per API key, and the bundled skills describe draft, review, publish. [source](https://docs.growthbook.io/integrations/mcp) |
| [Humblytics](https://duelpilot.com/compare/duelpilot-vs-humblytics) | Yes — Hosted MCP at mcp.humblytics.com (Streamable HTTP), included on every plan. [source 1](https://humblytics.com/skills), [source 2](https://mcp.humblytics.com/) | Partly — create\_split\_test drafts a test, but every write is a proposal the user must explicitly approve. [source](https://humblytics.com/skills) | No — No start tool; the page says nothing goes live until the user says so and the agent must never promote a winner itself. [source](https://humblytics.com/skills) |
| [Intelligems](https://duelpilot.com/compare/duelpilot-vs-intelligems) | Yes — Included in every paid plan; works with Claude, ChatGPT, Gemini and Grok. [source](https://www.intelligems.io/pricing) | Yes — create\_experience and update\_experience cover variations, targeting and traffic. [source](https://docs.intelligems.io/developer-resources/mcp-server/available-tools.md) | Yes — experience\_action can start, pause or end a test; new experiences are drafts and the AI client asks for confirmation before running them. [source](https://docs.intelligems.io/developer-resources/mcp-server/available-tools.md) |
| [Kameleoon](https://duelpilot.com/compare/duelpilot-vs-kameleoon) | Yes — mcp.kameleoon.com/mcp with OAuth; Claude, ChatGPT, Claude Code, Codex and Cursor. [source 1](https://docs.kameleoon.com/mcp/mcp-server), [source 2](https://docs.kameleoon.com/mcp/connect-ai-assistants) | Partly — experiment\_create makes a draft with common JS/CSS; no documented tool to create Web variation code. [source](https://docs.kameleoon.com/mcp/mcp-server) | Yes — experiment\_lifecycle\_update can start, pause, resume, stop or delete; only stop and delete require confirmation. [source](https://docs.kameleoon.com/mcp/mcp-server) |
| [LaunchDarkly](https://duelpilot.com/compare/duelpilot-vs-launchdarkly) | Yes — Hosted MCP at mcp.launchdarkly.com with OAuth. [source](https://launchdarkly.com/docs/home/getting-started/mcp-hosted) | Yes — create-flag, create-experiment, update-experiment. [source](https://launchdarkly.com/docs/home/getting-started/mcp-hosted) | Yes — Tools such as toggle-flag, update-rollout and start-experiment-iteration exist; where an environment requires approval, the agent can only file an approval request for a reviewer. [source](https://launchdarkly.com/docs/home/getting-started/mcp-hosted) |
| [Mida](https://duelpilot.com/compare/duelpilot-vs-mida) | Yes — mcp.mida.so/mcp with OAuth and dynamic client registration; launched May 2026. [source](https://www.mida.so/mcp) | Yes — Describe the test and Mida builds the variant, targeting, goal and preview link. [source](https://www.mida.so/changelog) | Yes — Tests are drafts until launched, but launching and serving the winner are MCP actions with no separate dashboard approval. [source](https://www.mida.so/mcp) |
| [Optimizely](https://duelpilot.com/compare/duelpilot-vs-optimizely) | Yes — Hosted remote MCP with Opti ID OAuth; requires Optimizely Agent Platform to be enabled. [source 1](https://docs.optimizely.com/experimentation-strategy/docs/optimizely-experimentation-mcp-server-overview), [source 2](https://docs.optimizely.com/experimentation-strategy/docs/optimizely-experimentation-mcp-server-quickstart) | Partly — Can create and update flags, experiments and audiences, but Web variation code (custom HTML, CSS, JS) is not supported through MCP. [source](https://docs.optimizely.com/experimentation-strategy/docs/optimizely-experimentation-mcp-server-overview) | Unknown — No documented start or publish action; the agent confirms in the client before committing changes. [source](https://docs.optimizely.com/experimentation-strategy/docs/optimizely-experimentation-mcp-server-example-prompts) |
| [pagent](https://duelpilot.com/compare/duelpilot-vs-pagent) | Yes — app.pagent.ai/api/mcp, Streamable HTTP with OAuth or a personal token; works with Claude, ChatGPT, Codex, Cursor and others. [source](https://pagent.ai/docs/guides/mcp) | Yes — The agent can capture a hypothesis, turn it into a test and work with pages, audiences and conversion goals. [source](https://pagent.ai/docs/guides/mcp) | Yes — With write scope granted at OAuth the agent can start, pause, stop, roll back and mark winners shipped; pagent adds no second confirmation and relies on the client's confirmation setting. [source](https://pagent.ai/docs/guides/mcp) |
| [PostHog Experiments](https://duelpilot.com/compare/duelpilot-vs-posthog) | Yes — Official PostHog MCP (US and EU). [source](https://posthog.com/docs/model-context-protocol/tools) | Yes — experiment-create, experiment-create-from-prompt, experiment-update. [source](https://posthog.com/docs/model-context-protocol/tools) | Yes — experiment-launch, -pause, -resume, -end and -ship-variant; approval is left to the MCP client, and write tools can be removed with read-only mode. [source 1](https://posthog.com/docs/model-context-protocol/tools), [source 2](https://posthog.com/docs/model-context-protocol/faq) |
| [Statsig](https://duelpilot.com/compare/duelpilot-vs-statsig) | Yes — Official MCP server with experiment, gate and API tools. [source](https://docs.statsig.com/integrations/mcp) | Yes — experiment\_create and gate\_create. [source](https://docs.statsig.com/integrations/mcp) | Unknown — No dedicated start or ship tool is documented; update and destructive tools need a confirmation parameter and review policies still apply. [source](https://docs.statsig.com/integrations/mcp) |
| [Tailor](https://duelpilot.com/compare/duelpilot-vs-tailor) | Yes — Tailor MCP server for Claude, Claude Code and internal agents (Pro and above). [source](https://www.tailorhq.ai/integrations/mcp) | Yes — Agents build experiments and variants: headlines, CTAs, images, full sections, layout and scripts. [source](https://www.tailorhq.ai/integrations/mcp) | Yes — Optional: "Launch can be open to agents or reserved for humans"; agents can start, pause and ramp experiments when allowed. [source](https://www.tailorhq.ai/integrations/mcp) |
| [Unbounce (Smart Traffic)](https://duelpilot.com/compare/duelpilot-vs-unbounce) | Yes — Included with every Unbounce plan. [source](https://unbounce.com/product/mcp-server/) | Yes — "Create variants, allocate traffic, start or stop tests, and deploy winning variants." [source](https://unbounce.com/product/mcp-server/) | Yes — The agent can start and stop tests and deploy winners; A/B testing needs the Experiment plan or higher. [source](https://unbounce.com/product/mcp-server/) |
| [VWO (Wingify)](https://duelpilot.com/compare/duelpilot-vs-vwo) | Yes — Web-testing MCP at mcp.wingify.ai (14 tools) and a separate FME MCP for feature flags. [source 1](https://help.wingify.com/hc/en-us/articles/58792565345305), [source 2](https://developers.wingify.com/v2/docs/fme-mcp-server) | Yes — Create A/B and split-URL campaigns, clone and update configuration (not MVT). [source](https://help.wingify.com/hc/en-us/articles/58792565345305) | Yes — With Publish/Admin permission an agent can start, pause or stop a campaign; no server-side approval step is documented. [source](https://help.wingify.com/hc/en-us/articles/58792565345305) |
| [ABConvert](https://duelpilot.com/compare/duelpilot-vs-abconvert) | Partly — In beta, opt-in via support, Growth plan and above. [source](https://docs.abconvert.io/mcp/overview) | Yes — Creates price, shipping, theme, template, URL, offer and checkout tests; visual-editor tests stay wizard-only. [source](https://docs.abconvert.io/mcp/overview) | Yes — Tools cover launch, pause, resume and end; drafts are created after the user confirms in chat. [source](https://docs.abconvert.io/mcp/tools) |
| [Optibase](https://duelpilot.com/compare/duelpilot-vs-optibase) | Partly — MCP on all plans, but read-only. [source](https://www.optibase.io/blog/optibase-mcp-ai-assistant-integration) | No — The MCP cannot create, edit or delete tests. [source](https://www.optibase.io/blog/optibase-mcp-ai-assistant-integration) | No — Read-only MCP. [source](https://www.optibase.io/blog/optibase-mcp-ai-assistant-integration) |
| [Webflow Optimize](https://duelpilot.com/compare/duelpilot-vs-webflow-optimize) | Partly — Webflow's MCP can read Optimize experiments and results (announced 15 Oct 2026); no write access. [source](https://webflow.com/updates/mcp-optimize-analyze-aeo) | No — Only read access is announced. [source](https://webflow.com/updates/mcp-optimize-analyze-aeo) | No — People create and launch in the Webflow UI. [source](https://webflow.com/updates/mcp-optimize-analyze-aeo) |
| [AB Tasty](https://duelpilot.com/compare/duelpilot-vs-ab-tasty) | Unknown — No AB Tasty-specific MCP found; whether Wingify's MCP serves AB Tasty-platform accounts is not documented. [source](https://help.wingify.com/hc/en-us/articles/58792565345305) | Unknown — Not documented for the AB Tasty platform. [source](https://docs.abtasty.com/) | Unknown — Not documented for the AB Tasty platform. [source](https://docs.abtasty.com/) |
| [Fibr AI](https://duelpilot.com/compare/duelpilot-vs-fibr) | Unknown — No MCP server or public API mentioned. [source](https://fibr.ai) | Unknown — Fibr's own AI generates variants; no route for a customer's agent documented. [source](https://fibr.ai/experimentation) | Unknown — Not documented for external agents; in the platform, variants are approved with one click and winners can be auto-promoted. [source](https://fibr.ai/experimentation) |
| [Coframe](https://duelpilot.com/compare/duelpilot-vs-coframe) | No — No customer MCP server for test management found; the site only exposes a WebMCP recommendation tool for visitors' agents. [source](https://www.coframe.com) | No — Coframe's own AI creates variants; your own agent cannot. [source](https://www.coframe.com) | No — Variations need the customer's approval before going live. [source](https://www.coframe.com) |

## Connect Duelpilot's MCP server

The server is at https://duelpilot.com/api/mcp (Streamable HTTP, stateless). Pick what your client supports:

- claude.ai, Claude Desktop and ChatGPT: add a custom connector with URL https://duelpilot.com/api/mcp and sign in (OAuth). You pick the organisation, sites and permissions; there is no token to paste.
- Claude Code: run the command below with your Duelpilot token (the tools appear in a new session).
- Cursor and other MCP clients: a remote server with URL https://duelpilot.com/api/mcp and the header Authorization: Bearer <token>.
- No account yet? Your agent can open a workspace itself and hand you a claim link: [instructions for agents](https://duelpilot.com/agents.md).

```bash
claude mcp add --transport http duelpilot https://duelpilot.com/api/mcp --header "Authorization: Bearer $DUELPILOT_TOKEN"
```

## What your agent can and can't do

48 tools cover installing and verifying the script, reading GA4 and page structure, proposing ideas, designing variants with previews and device checks, reading results and posting updates. Its only protective action is pausing a test. There is deliberately no tool that approves, launches, resumes, picks a winner, ships, changes policy or manages tokens: those stay with people in Duelpilot's board.

## Try Duelpilot with your own agent

Give your agent one line: [set up Duelpilot for my site using /agents.md](https://duelpilot.com/agents.md). It opens a workspace, installs and verifies the script, and drafts the first test; you approve what goes live. [Try your first test free](https://duelpilot.com/login) or [see pricing](https://duelpilot.com/#pricing).

## FAQ

### What is an A/B testing MCP server?

A Model Context Protocol server that exposes an A/B testing tool's actions (read results, propose a test, check a page) so an AI agent like Claude, ChatGPT or Cursor can use the tool directly.

### Which A/B testing tools have an MCP server?

Duelpilot, ABConvert, Amplitude Experiment, Convert, Crazy Egg, Eppo (Datadog Experiments), Flint, GrowthBook, Humblytics, Intelligems, Kameleoon, LaunchDarkly, Mida, Optibase, Optimizely, PostHog Experiments, Statsig, Tailor, Unbounce (Smart Traffic), VWO (Wingify), Webflow Optimize, pagent, as documented by each vendor (some only partly; see the table).

### Can an agent launch tests through Duelpilot's MCP server?

No. Agents propose and check; a person approves every launch in Duelpilot's board (or an owner's autopilot policy within guardrails).

## Sources

- [Anthropic: Introducing the Model Context Protocol (25 Nov 2024)](https://www.anthropic.com/news/model-context-protocol) (checked 2026-10-17)
- [Agent-Native CRO for Claude & Codex | Humblytics](https://humblytics.com/skills) (checked 2026-10-17)
- [Humblytics MCP](https://mcp.humblytics.com/) (checked 2026-10-17)
- [Pagent MCP server for Claude, ChatGPT, Codex and Cursor](https://pagent.ai/docs/guides/mcp) (checked 2026-10-17)
- [Tailor MCP: Connect Your AI Agents to Website Experiments](https://www.tailorhq.ai/integrations/mcp) (checked 2026-10-17)
- [Coframe — AI landing pages that maximize revenue](https://www.coframe.com) (checked 2026-10-17)
- [Optimizely Experimentation MCP server overview](https://docs.optimizely.com/experimentation-strategy/docs/optimizely-experimentation-mcp-server-overview) (checked 2026-10-17)
- [Optimizely Experimentation MCP server quickstart](https://docs.optimizely.com/experimentation-strategy/docs/optimizely-experimentation-mcp-server-quickstart) (checked 2026-10-17)
- [Optimizely Experimentation MCP server example prompts](https://docs.optimizely.com/experimentation-strategy/docs/optimizely-experimentation-mcp-server-example-prompts) (checked 2026-10-17)
- [Kameleoon MCP server](https://docs.kameleoon.com/mcp/mcp-server) (checked 2026-10-17)
- [Connect AI assistants | Kameleoon](https://docs.kameleoon.com/mcp/connect-ai-assistants) (checked 2026-10-17)
- [Mida MCP server | Connect Mida to your AI agent](https://www.mida.so/mcp) (checked 2026-10-17)
- [Changelog | Mida](https://www.mida.so/changelog) (checked 2026-10-17)
- [Fibr AI](https://fibr.ai) (checked 2026-10-17)
- [Fibr AI Experimentation](https://fibr.ai/experimentation) (checked 2026-10-17)
- [Webflow updates: MCP for Optimize, Analyze and AEO](https://webflow.com/updates/mcp-optimize-analyze-aeo) (checked 2026-10-17)
- [Convert MCP Server](https://www.convert.com/mcp-server/) (checked 2026-10-17)
- [Convert MCP Server | Convert Support](https://support.convert.com/hc/en-us/articles/convert-mcp-server) (checked 2026-10-17)
- [Connect Wingify with AI Tools using Wingify's MCP Server](https://help.wingify.com/hc/en-us/articles/58792565345305) (checked 2026-10-17)
- [MCP Server | Wingify FME](https://developers.wingify.com/v2/docs/fme-mcp-server) (checked 2026-10-17)
- [AB Tasty Docs](https://docs.abtasty.com/) (checked 2026-10-17)
- [PostHog MCP tools reference](https://posthog.com/docs/model-context-protocol/tools) (checked 2026-10-17)
- [PostHog MCP FAQ](https://posthog.com/docs/model-context-protocol/faq) (checked 2026-10-17)
- [Pricing & Plans | Intelligems](https://www.intelligems.io/pricing) (checked 2026-10-17)
- [Available Tools | Intelligems Docs](https://docs.intelligems.io/developer-resources/mcp-server/available-tools.md) (checked 2026-10-17)
- [Drive ABConvert from Claude with MCP](https://docs.abconvert.io/mcp/overview) (checked 2026-10-17)
- [What Claude can do for you | ABConvert Docs](https://docs.abconvert.io/mcp/tools) (checked 2026-10-17)
- [Optibase MCP: Ask AI About Your A/B Tests in Plain English](https://www.optibase.io/blog/optibase-mcp-ai-assistant-integration) (checked 2026-10-17)
- [Crazy Egg Model Context Protocol (MCP) Connector](https://support.crazyegg.com/knowledge-base/crazy-egg-model-context-protocol-mcp/) (checked 2026-10-17)
- [Unbounce MCP: Build Landing Pages in Claude & ChatGPT](https://unbounce.com/product/mcp-server/) (checked 2026-10-17)
- [MCP Server | GrowthBook Docs](https://docs.growthbook.io/integrations/mcp) (checked 2026-10-17)
- [MCP | Statsig Docs](https://docs.statsig.com/integrations/mcp) (checked 2026-10-17)
- [Datadog MCP Server Tools](https://docs.datadoghq.com/mcp_server/tools/) (checked 2026-10-17)
- [Amplitude MCP](https://amplitude.com/docs/amplitude-ai/amplitude-mcp) (checked 2026-10-17)
- [LaunchDarkly hosted MCP server](https://launchdarkly.com/docs/home/getting-started/mcp-hosted) (checked 2026-10-17)
- [Claude MCP | Flint Docs](https://www.flint.com/docs/claude-mcp) (checked 2026-10-17)
- [Flint MCP | Connect Claude to Your Landing Page Workflow](https://www.flint.com/product/flint-mcp) (checked 2026-10-17)
- [Duelpilot instructions for AI agents](https://duelpilot.com/agents.md) (checked 2026-10-17)
- [Duelpilot overview](https://duelpilot.com/index.md) (checked 2026-10-17)
- [Duelpilot pricing](https://duelpilot.com/pricing.md) (checked 2026-10-17)

## Related

- [Agentic A/B testing explained](https://duelpilot.com/agentic-ab-testing)
- [Best A/B testing tools in 2026](https://duelpilot.com/best-ab-testing-tools)
- [All comparisons](https://duelpilot.com/compare)