Adrails

What the Adrails MCP server does

The Adrails MCP server gives supported AI clients access to paid-media and business data held in an Adrails workspace. It connects Meta Ads and Google Ads accounts with other configured sources, including Shopify, Klaviyo, Stripe, Triple Whale, and PostHog. Depending on the permissions granted and the user’s workspace role, an agent can inspect advertising performance, compare periods, review business results, and work with campaign planning data.

The server also supports campaign-related tasks such as reviewing structures, drafts, templates, media, audiences, and keywords. Google Keyword Planner volumes are included in the available research capabilities. Workspace tools cover automations, Knowledge, channels, and job status.

How it works

Adrails hosts the MCP endpoint at https://adrails.ai/api/mcp. Clients connect over Streamable HTTP and authenticate with OAuth 2.1 using PKCE and dynamic client registration. During sign-in, the user selects an Adrails workspace and grants the permissions available on the consent screen.

The server exposes 63 tools, with the actual set filtered by workspace permissions and role. Tool calls do not use an Adrails AI model and therefore do not consume agent credit, except when an automation includes an agent step.

Campaign changes follow a proposal workflow. The agent can prepare budget, status, or launch changes and return a link showing the precise before-and-after state. The change is not written to a Meta or Google Ads account automatically; the user applies it in Adrails while signed in, before the proposal expires.

Setup and configuration

The Adrails MCP server requires an Adrails account, but no package installation, local process, or environment variable is described for this repository. Add the shared server URL to an MCP client, let the client register, and complete the Adrails login and consent flow.

Claude users can add a custom connector using the server URL and leave the OAuth client ID and secret blank. Claude Code supports the claude mcp add --transport http adrails https://adrails.ai/api/mcp command. Codex can be configured with codex mcp add adrails --url https://adrails.ai/api/mcp, followed by codex mcp login adrails. Other client instructions are provided in Adrails’ connection documentation.

Tools and capabilities

The Adrails MCP server organizes its tools around several areas:

• Performance by period or day, metric comparisons, diagnosis, portfolio reviews, and business reviews across Meta and Google Ads.
• Store revenue alongside advertising metrics, plus orders, customers, and revenue from connected business systems.
• Campaign structures, drafts, templates, media, audience research, and keyword research.
• Preparation of budget, status, and campaign-launch changes with an exact before-and-after link.
• Workspace automations, Knowledge, channels, and job-status information.

Access to each tool depends on the permissions granted during connection and the user’s workspace role.

Limitations and notes

The repository contains listing metadata rather than a server implementation. Adrails hosts the service, so there is nothing to run locally. The README does not describe local configuration, environment variables, or a self-hosted mode.

Connections can be revoked from the MCP page in Adrails. Although agents can prepare advertising changes, the server does not apply those changes by itself. Users must review and apply proposals in Adrails, and proposals are subject to expiration. Adrails is not affiliated with Meta Platforms or Google LLC.