mcpField
A
AdChat

AdChat

Remote
adchat.appBusiness & Marketing1Updated 1d ago

About AdChat

AdChat is a hosted MCP server for managing and analyzing Facebook and Instagram ads through Meta's Marketing API. It gives AI app users access to live account reporting, with campaign changes available when write access is enabled.

Key features

  1. 01Query live ad performance
  2. 02Review account and campaign details
  3. 03Create and edit campaigns, ad sets, and ads
  4. 04Research audiences and ad creatives
  5. 05Review, approve, reject, or undo changes

Use cases

  1. 01Compare ad results across a reporting period
  2. 02Break down spend by audience or placement
  3. 03Draft a budget change for approval
  4. 04Research audiences before campaign setup

MCP support

AdChat is a remote MCP server using Streamable HTTP and OAuth 2.1 with PKCE. It exposes read tools by default; write tools require write access to be enabled in the AdChat dashboard.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://mcp.adchat.app
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

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 README

AdChat: Meta Ads MCP server

Run Facebook and Instagram ads from your AI app. AdChat is a hosted, remote Model Context Protocol server for the Meta Marketing API: ask how your ads are doing in plain English, get answers from your live numbers, and (when you turn it on) create, edit, pause and resume campaigns, ad sets and ads.

https://mcp.adchat.app
  • Transport: Streamable HTTP
  • Auth: OAuth 2.1 with PKCE (S256) and dynamic client registration. Sign in with your AdChat account; no API keys, no Meta tokens to copy.
  • Works in: Claude, ChatGPT, Grok and Grok Bot, Cursor, VS Code, Claude Code, Codex, and any app that accepts a remote MCP server URL.
  • Website: https://adchat.app · Setup for every app: https://adchat.app/connect

This repository holds documentation and directory metadata for the hosted server. The server itself is operated by AdChat and is not open source.

Connect

AppHow
Claude (claude.ai, desktop)Customize → Connectors → + Add → Add custom connector, name it AdChat, paste the URL. Guide
ChatGPTchatgpt.com/plugins → + → Add custom MCP server, paste the URL, choose OAuth. Guide
Grokgrok.com/connectors → New Connector → Custom, paste the URL. Guide
Grok BotAsk your bot: Add a custom MCP server called adchat at https://mcp.adchat.app and start the sign-in.
Claude Codeclaude mcp add --transport http adchat https://mcp.adchat.app, then /mcp to sign in
CursorAdd to Cursor, or add to ~/.cursor/mcp.json (below)
VS CodeAdd to VS Code, or add an HTTP MCP server with the URL
Anything elsePaste the URL where the app asks for a remote MCP server and sign in when prompted
{
  "mcpServers": {
    "adchat": { "url": "https://mcp.adchat.app" }
  }
}

The first time you connect, you sign in to AdChat (free, no card), connect Facebook once, and choose which ad accounts the AI may see.

What you can ask

  • "How did my Facebook ads do last week? Which ad had the lowest cost per result?"
  • "Break down last month's spend by age, placement and country."
  • "Which ad sets are spending without results?"
  • "Draft a 15% budget increase on my best ad set and show me before you apply it."
  • "Build a 1% lookalike of my purchasers in the US and estimate its size."

Tools

55 tools on the hosted server. Read tools are always on; write tools need write access turned on in the AdChat dashboard.

Accounts and reporting: get_ad_accounts, get_account_info, get_account_pages, get_account_briefing, get_insights, get_true_profit, get_alerts

Campaigns, ad sets, ads: get_campaigns, get_campaign_details, create_campaign, update_campaign, get_adsets, get_adset_details, create_adset, update_adset, create_budget_schedule, get_ads, get_ad_details, create_ad, update_ad

Creatives and media: get_ad_creatives, get_creative_details, create_ad_creative, update_ad_creative, get_ad_image, get_image_by_hash, get_ad_video, upload_ad_image, get_upload_link, compute_image_crops

Targeting and audiences: search_interests, get_interest_suggestions, search_behaviors, search_demographics, search_geo_locations, estimate_audience_size, get_custom_audiences, get_custom_audience_details, create_website_audience, create_engagement_audience, create_lookalike_audience, delete_custom_audience

Research: search_ads_archive (Meta Ad Library), search_pages_by_name

Changes and safety: list_pending_changes, approve_change, reject_change, undo_change, get_change_history

Memory and files: remember, save_file, list_files, read_file, ask_user, get_login_link

Safe by default

  • Read-only until you say otherwise. Write tools are refused until write access is turned on for your account.
  • Paused by default. New campaigns, ad sets and ads are created paused, so nothing spends by surprise.
  • Drafts and approval (Pro). Changes can be returned as drafts you approve first, checked against spend rules (for example a cap on budget increases per 24 hours) and protected campaigns that are enforced in code.
  • Undo (Pro). Most changes can be reversed in one click for 30 days.
  • Account allowlist. The AI only sees the ad accounts you picked.
  • Rate limit: 120 requests per minute per user.

AdChat is an approved Meta app with Standard Access to the Marketing API, so it works on any ad account you manage in Ads Manager.

Pricing

Free plan: read-only, one ad account, no card. Paid plans: Connect $29/month, Pro $49/month, Agency $99/month. Current details: https://adchat.app/pricing

Links

Listed in the official MCP Registry as app.adchat/meta-ads.

Comments

Markdown supported

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.