About Unfetch
Unfetch is an agent plugin for reporting on connected marketing accounts, turning live data into reports and follow-up analysis. It covers Google and Meta advertising, analytics, Search Console, and Merchant Center, with read-only access by default.
Key features
- 01Google Ads campaign and performance reports
- 02Meta Ads reporting across placements
- 03Analytics and Search Console reports
- 04Merchant Center product insights
- 05Keyword research and web research
- 06Private reporting preferences
Use cases
- 01Reviewing campaign spend and conversions
- 02Investigating search queries and landing pages
- 03Checking product performance and pricing benchmarks
- 04Creating reports with business-specific preferences
MCP support
Unfetch provides an MCP connection using a production Streamable HTTP endpoint, with OAuth authorization. It connects supported MCP clients to data sources configured in Unfetch.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://unfetch.com/api/mcpclaude mcp add --transport http com-unfetch-unfetch https://unfetch.com/api/mcp{"mcpServers":{"com-unfetch-unfetch":{"type":"http","url":"https://unfetch.com/api/mcp"}}}Open in Cursor{"servers":{"com-unfetch-unfetch":{"type":"http","url":"https://unfetch.com/api/mcp"}}}Open in VS Code{"mcpServers":{"com-unfetch-unfetch":{"type":"http","url":"https://unfetch.com/api/mcp"}}}{"mcpServers":{"com-unfetch-unfetch":{"serverUrl":"https://unfetch.com/api/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEUnfetch Agent Plugin
Connect your AI assistant to your marketing data and turn live account evidence into clear, actionable reports.
Unfetch supports:
- Google Ads campaign, keyword, search-term, conversion, budget, and performance reporting
- Meta Ads campaign, ad set, ad, placement, spend, conversion, and performance reporting across Facebook, Instagram, and Messenger
- Google Analytics traffic and conversion reporting
- Google Search Console query and landing-page reporting
- Google Merchant Center product and account issues, free-listing and Shopping product performance, price benchmarks and suggestions, best sellers, and Shopping competitor visibility
- Keyword research, including demand, competition, CPC, and bid estimates
- Web research for current public-page and landing-page evidence
- Follow-up analysis across report results
- Private memories for reporting preferences and business context
Access is read-only by default: a newly connected AI client cannot change campaigns, budgets, bids, ads, analytics properties, Search Console data, or Merchant Center products. In AI client settings you can allow a client to propose account changes, which you approve or reject in Unfetch, or to apply changes directly. Account changes also require admin access to the brand and a paid plan. Personal memory tools can save or delete private reporting preferences in your Unfetch account.
Before you install
- Sign in to Unfetch.
- Create or select a brand, then open Integrations in that brand's navigation to connect its data sources. Create a brand if you do not have one yet.
- Confirm that your team membership and brand access include every brand you want the assistant to use.
OAuth opens automatically when a supported client connects. Access always follows your current Unfetch team membership and brand assignments.
Install in ChatGPT desktop
A ChatGPT workspace admin can import Unfetch from GitHub:
- Open Admin > Plugins in ChatGPT desktop.
- Select Add, then Import marketplace.
- Enter
https://github.com/unfetch-com/agent-pluginas the Source. Leave Path and Branch, tag, or commit empty. - Import the marketplace, review Unfetch, and make it available in the workspace.
- Open the Plugins tab and install Unfetch.
The installation includes the reporting skill and MCP connection. Direct GitHub marketplace imports require a ChatGPT workspace admin.
Install in Claude
Add Unfetch once to your Claude account:
- Claude web: open Unfetch in Claude's plugin directory, select Add, then complete authorization when prompted.
- Claude Desktop: open Customize > Plugins > Discover, search for Unfetch, select it, then click Add.
- Claude Code: sign in with the same Claude account. Restart Claude Code or run
/reload-pluginsif Unfetch is not available immediately.
If account sync is unavailable, install directly from GitHub by running these commands inside Claude Code:
/plugin marketplace add unfetch-com/agent-plugin
/plugin install unfetch@unfetch-plugins
Install in Codex
Run:
codex plugin marketplace add https://github.com/unfetch-com/agent-plugin
codex plugin add unfetch@unfetch-plugins
Start a new Codex session after installation if the plugin is not available in the current session.
Install in Gemini CLI
Run:
gemini extensions install https://github.com/unfetch-com/agent-plugin
Restart Gemini CLI after installation if the extension is not available in the current session.
Install in VS Code with GitHub Copilot
- Open the Command Palette.
- Run Chat: Install Plugin From Source.
- Enter
https://github.com/unfetch-com/agent-plugin. - Review the source and confirm the installation.
You can also open Chat: Open Customizations, select Plugins, and choose Install Plugin from Source.
ChatGPT web and other MCP clients
Use the guided instructions on the Unfetch plugin setup page. It provides the production Streamable HTTP connection and handles OAuth without API keys or credentials in this repository.
Access and privacy
- Manage, limit, or revoke connected AI clients and their account access from AI client settings.
- Manage members and brand access from Team settings.
- Manage data-source connections from Integrations inside each brand.
- Review the privacy policy and terms of service.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.