Meta Ads (Pipeboard) MCP Server
Manage Facebook & Instagram ad campaigns from your AI client via the Meta Ads (Pipeboard) MCP server.
Updated
What is Meta Ads (Pipeboard) MCP?
The Meta Ads MCP server is a Model Context Protocol server that lets AI assistants and agents create, manage, and analyze Facebook and Instagram (Meta) advertising campaigns through a standard tool interface.
Built and hosted by Pipeboard, it is a remote, cloud-hosted server — there is no local Node or Python process to run. You point your MCP client at the server URL (https://mcp.pipeboard.co/meta-ads-mcp) and authenticate via OAuth, connecting your Meta ad accounts once at pipeboard.co. Pipeboard handles the OAuth handshake with Meta, so your AD-platform credentials are never exposed to the AI client.
Once connected, the server exposes the Meta Marketing API's object hierarchy — account → campaign → ad set → ad → creative — as agent-callable tools, covering everything from retrieving performance insights to creating campaigns, uploading creatives, and searching targeting options. Write operations (like creating campaigns) default to a PAUSED state for safety.
Tools the Meta Ads (Pipeboard) MCP server exposes
| Tool | What it does |
|---|---|
| mcp_meta_ads_get_ad_accounts | Retrieve the ad accounts accessible to the connected user. |
| mcp_meta_ads_get_account_info | Fetch detailed information about a specific ad account. |
| mcp_meta_ads_get_campaigns | List campaigns, optionally filtered by status. |
| mcp_meta_ads_create_campaign | Create a new campaign with objectives and budgets (defaults to PAUSED). |
| mcp_meta_ads_get_adsets | Retrieve ad sets, optionally filtered by campaign. |
| mcp_meta_ads_create_adset | Create a new ad set with targeting and budget settings. |
| mcp_meta_ads_create_ad | Launch a new ad using an existing creative. |
| mcp_meta_ads_create_ad_creative | Build a new ad creative from an uploaded image. |
| mcp_meta_ads_upload_ad_image | Upload an image file for use in ad creatives. |
| mcp_meta_ads_get_insights | Retrieve campaign/ad performance metrics with optional breakdowns. |
Connect the Meta Ads (Pipeboard) MCP server
Claude Code
- 1
Add the remote MCP server
Run: claude mcp add --transport http meta-ads https://mcp.pipeboard.co/meta-ads-mcp
- 2
Authenticate via OAuth
On first use, Claude Code opens the Pipeboard OAuth flow. Sign in and connect your Meta ad account(s) at pipeboard.co. Alternatively append ?token=YOUR_PIPEBOARD_TOKEN to the URL using a token from pipeboard.co/api-tokens.
- 3
Verify the connection
Run /mcp in Claude Code to confirm the meta-ads server is connected and its tools are listed.
Cursor
- 1
Edit your MCP config
Open ~/.cursor/mcp.json (or the project .cursor/mcp.json).
- 2
Add the server entry
Add: {"mcpServers": {"meta-ads": {"url": "https://mcp.pipeboard.co/meta-ads-mcp"}}}
- 3
Authenticate and enable
Reload Cursor, complete the Pipeboard OAuth prompt (connect your Meta ad account at pipeboard.co), then enable the meta-ads server in Settings → MCP.
Security & permissions
Authentication uses OAuth 2.0 through Pipeboard. You connect your Meta ad accounts once at pipeboard.co; Pipeboard states it never sees or stores your Meta credentials. The MCP client authenticates to the server with a Pipeboard token, and the agent is granted access to the Meta ad accounts you authorize — including the ability to read insights and account data and to create or modify campaigns, ad sets, ads, and creatives. Because write tools can change live ad spend, scope access carefully and require human approval for write actions. New campaigns are created in a PAUSED state by default.
Using Meta Ads (Pipeboard) MCP with Gamut
In a Gamut workflow, the Meta Ads MCP server runs as a connected tool that an agent calls on a schedule or in response to an event — Gamut manages the OAuth connection so the agent can act without per-run sign-in. A scheduled agent might pull get_insights each morning to summarize spend and ROAS across campaigns, flag underperforming ad sets, and post the digest to a chat channel; an event-triggered agent might react to a budget-threshold alert by reviewing campaign details and (optionally) pausing or adjusting an ad set. Because write actions can create or change live spend, gate them behind explicit approval steps in the workflow.
Frequently asked questions
What is the Meta Ads MCP server?
It is a Model Context Protocol server, built by Pipeboard, that lets AI assistants and agents manage Facebook and Instagram ad campaigns — creating and updating campaigns, ad sets, ads, and creatives, and pulling performance insights — by exposing the Meta Marketing API as agent-callable tools.
Is the Meta Ads MCP server free?
Pipeboard offers a free plan with no credit card required to get started, plus paid tiers listed at pipeboard.co/pricing. Note that running ads on Meta itself incurs separate ad spend billed by Meta.
How do I connect the Meta Ads MCP server to Claude or Cursor?
Point your client at the remote URL https://mcp.pipeboard.co/meta-ads-mcp and complete the Pipeboard OAuth flow. In Claude Code use 'claude mcp add --transport http meta-ads <url>'; in Cursor add a url entry to mcp.json. Connect your Meta ad accounts once at pipeboard.co.
Is the Meta Ads MCP server hosted or self-hosted?
It is remote and cloud-hosted by Pipeboard, so there's no local install — you just configure the server URL and authenticate via OAuth. The open-source code is also available on GitHub (pipeboard-co/meta-ads-mcp) if you prefer to run it yourself with a local transport.