Workmagic Mcp

io.github.workmagicioPublished by GitHub account workmagiciov1.0.2Updated Nov 11, 2026

Marketing attribution, incrementality and budget planning from your WorkMagic account.

VerifiedStreamable HTTPWeb executableData & AnalyticsBusiness & Commerce

Overview

AI-generated overview

Connects an AI assistant to a WorkMagic account to query marketing attribution, incrementality test results and media budget plans in plain language.

What you can do
  • Query attribution metrics such as ROAS, CAC, revenue and orders by channel, campaign or period
  • Compare attribution models (iDDA, DDA, last-click, platform-reported) and explain differences
  • Diagnose attribution drops, spikes or zeros, and break results down by custom dimensions
  • Interpret completed lift tests and read, compare or reconcile Media Budget Optimizer scenarios
  • Create or edit a budget scenario from a plain-language budget question
When to use it
Use it when you run paid marketing on channels like Meta, Google or TikTok and want an assistant to answer questions about attribution, incrementality and budget allocation from your own WorkMagic data. It suits ecommerce and DTC teams already using WorkMagic who want conversational analysis instead of dashboards.
Requirements
A remote Streamable HTTP MCP server at mcp.workmagic.io, added as a custom connector or remote server. An active WorkMagic account with OAuth sign-in on first use; lift test and budget optimizer features must be enabled on the account. No packages or environment variables are needed.
Before you install
Sign-in grants the server access to your WorkMagic account data, and requests are logged by WorkMagic for security and support. Some skills can change account settings, such as creating or editing a budget scenario, so confirm proposed changes before accepting them. Lift test and budget optimizer features require an eligible paid plan.

Installation

In SourceWeft

  1. Open Workmagic Mcp in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/io.github.workmagicio%2Fworkmagic-mcp/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "workmagic": {
      "type": "http",
      "url": "https://mcp.workmagic.io/mcp"
    }
  }
}

README

WorkMagic for AI assistants

[WorkMagic for AI assistants — multi-touch attribution, incrementality and MMM in Claude, ChatGPT, Codex, Cursor, Copilot and Gemini]

Ask your marketing measurement data anything — in Claude, ChatGPT, Codex, Cursor, Copilot or Gemini. Attribution by channel and campaign, why a number moved, lift test results and media budget plans, answered from your own WorkMagic account.

[Add to Claude] [ChatGPT] [Codex] [Cursor] [Gemini CLI] [MCP] [License: MIT] [Follow on LinkedIn]

⭐ Star this repo to follow new skills and platforms as they ship — see the changelog.

What is WorkMagic

WorkMagic is an AI-powered marketing measurement platform for ecommerce and DTC brands. It brings multi-touch attribution (MTA), incrementality (geo lift tests) and media mix modeling (MMM) / budget planning together — across Meta, Google, TikTok and more — so you can see what's actually driving sales and decide where to spend next.

This open-source (MIT) plugin lets your AI assistant ask that same data questions in plain language, using your own WorkMagic account — an open-source MCP server plus skills for marketing attribution, incrementality and MMM.

See it in action

You: Compare Meta and Google attributed revenue under data-driven, last click, first click and any-click attribution for the last 30 days, and explain why they differ.

Claude: (abridged — WorkMagic demo store, synthetic data)

ChannelData-drivenLast clickFirst clickAny click
Meta$10.2K$17.3K$27.1K$38.0K
Google$12.7K$19.9K—$33.2K

Meta looks strongest under first- and any-click because it starts most journeys: shoppers first arrive from a Meta ad, then come back through Google or direct before buying. Data-driven attribution splits each order's credit across those touchpoints, so both channels shrink. Any-click gives every touched channel full credit, which is why the two any-click figures add up to more than the store actually sold ($46.9K).

More sample conversations: readiness check · model comparison · best-selling products

What's included

The plugin has two parts: the WorkMagic MCP server (https://mcp.workmagic.io/mcp), which reads and updates your WorkMagic account after you sign in, and nine skills that tell the assistant how to use it well.

SkillUse it when you want to
attribution-data-queryGet attribution numbers — ROAS, CAC, revenue, orders by channel, campaign or period
attribution-intent-clarificationAsk a vague question; the assistant asks one focused follow-up, then answers
attribution-model-comparisonSee iDDA, DDA, last-click and platform-reported numbers side by side, and why they differ
attribution-anomaly-diagnosisUnderstand why attribution dropped, spiked or shows zero
attribution-custom-dimensionBreak results down by audience, region, brand or another label in your campaign names
attribution-edge-routingGet pointed to the right place when a question is outside attribution
lift-test-readoutInterpret a completed lift test (incremental ROAS, confidence) and decide what to do next
mbo-read-scenarioRead, compare or reconcile a Media Budget Optimizer scenario
mbo-create-scenarioAsk "how should I split my budget?" and get a configured MBO scenario

Try asking:

  • "What was my Meta ROAS last week?"
  • "Why did Google attribution drop on Tuesday?"
  • "Which products sold the most in the last 30 days?"
  • "How should I split a $200K budget next month?"
  • "Is my account ready to run a Meta lift test?"

Get WorkMagic

Using this plugin needs an active WorkMagic account.

  • New to WorkMagic? See plans and pricing to get started.
  • Already a customer? Install below and sign in with your WorkMagic account the first time the assistant uses it.

Questions about plans? Contact your WorkMagic CSM or [email protected]. Lift test and budget optimizer skills need those features enabled on your account.

Install

Claude (claude.ai, desktop, Cowork)

  1. Open plugin settings (Settings → Plugins, or Plugins & skills), choose Add marketplace → GitHub repo, enter workmagicio/workmagic-agent-plugin, then install WorkMagic. (Listing in the Claude plugin directory is coming soon.)
  2. Open the plugin's Connectors tab and connect WorkMagic.

Only want the data connection? Add https://mcp.workmagic.io/mcp under Customize > Connectors as a custom connector.

Claude Code

/plugin marketplace add workmagicio/workmagic-agent-plugin/plugin install workmagic@workmagic-plugins

Codex

codex plugin marketplace add workmagicio/workmagic-agent-plugin

Then install workmagic from /plugins.

本地安装:下载 release 中的 workmagic-codex.zip,或在仓库中执行以下命令构建并安装:

sh
python3 tools/build_zip.py codexunzip dist/workmagic-codex.zip -d dist/workmagic-codexcodex plugin marketplace add ./dist/workmagic-codexcodex plugin add workmagic@workmagic-plugins

workmagic-chatgpt.zip 用于 ChatGPT 插件门户;Codex 本地安装使用 workmagic-codex.zip。

Gemini CLI

gemini extensions install https://github.com/workmagicio/workmagic-agent-plugin

ChatGPT, Cursor, GitHub Copilot

Directory listings are coming soon. Until then, connect the MCP server below; in Cursor you can also add this repository as a team marketplace.

Any other MCP client

Add https://mcp.workmagic.io/mcp as a remote (Streamable HTTP) MCP server. You get the tools; the skills in skills/ are plain Markdown you can load as instructions.

The first time the assistant uses WorkMagic, sign in with your WorkMagic account. If you belong to several WorkMagic accounts, the assistant asks which one to use. Don't have an account yet? See Get WorkMagic above.

Data

The skills are instructions only; they don't run code or contact any other service. The assistant reaches your data through the WorkMagic MCP server at https://mcp.workmagic.io/mcp, using the account you sign in with. The server receives the requests the assistant makes on your behalf, such as queries, filters and settings you ask to change, and returns your WorkMagic data. WorkMagic logs those requests for security and support. Nothing is sent anywhere else.

Some skills can change settings in your WorkMagic account, such as creating or editing a budget scenario. The assistant shows what it will change and asks you to confirm first.

The tools/ folder holds scripts WorkMagic uses to maintain this repository. The plugin never runs them.

Follow WorkMagic

New skills, platforms and product updates ship often — follow along:

Support

Contact your WorkMagic CSM or [email protected]. Privacy policy: www.workmagic.io/privacy. Found a problem with a skill? Open an issue.

Source: README.md at commit 24e5145

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

2
  1. v1.0.2LatestNov 11, 2026
  2. v1.0.1Sep 16, 2026

Score

85% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts