paolobietolini/gtm-mcp-server

Description and use

**Brief Summary (2‑3 sentences)**
paolobietolini/gtm‑mcp‑server is an open‑source Model‑Context‑Protocol (MCP) server that lets you control Google Tag Manager (GTM) with natural‑language prompts via any LLM. After a one‑time authentication step, the server exposes a simple API/CLI that AI agents can call to create, update, or query GTM containers, tags, triggers, and variables.

**Value**

• **Bridges the AI‑to‑tool gap**: By translating natural‑language intents into concrete GTM actions, the server enables conversational AI assistants to manage marketing tags without writing code.
• **Standardized integration**: It implements the MCP spec, so the same server can be swapped with other MCP‑compatible tools, reducing vendor lock‑in and simplifying multi‑tool orchestration.
• **Security‑first workflow**: Authentication is performed once, after which the server enforces scoped permissions, keeping GTM credentials safe while still permitting automated updates.

**Practical Adoption Path**

1. **Deploy the server** – Run the Go binary (Docker image is also provided) in your cloud or on‑prem environment.
2. **Authenticate GTM** – Follow the built‑in OAuth flow to grant the server the required GTM scopes; the resulting token is stored securely.
3. **Connect your LLM** – Configure your LLM (e.g., OpenAI, Anthropic) to use the server’s MCP endpoint as a tool plugin or via the provided SDK/CLI.
4. **Define intents** – Map natural‑language commands (e.g., “Add a GA4 tag to the homepage”) to MCP calls; the server handles the translation to GTM’s REST API.
5. **Iterate & monitor** – Use the server’s logs and optional webhook callbacks to audit changes and refine prompt handling.

**Production Readiness**

• **Activity & community**: 83 ★, 18 forks, recent commit (2026‑05‑12), and a Go codebase with clear module structure indicate an active project.
• **Stability**: The MCP implementation is self‑contained, exposing a well‑documented HTTP/JSON interface; the Go runtime provides strong concurrency and low overhead.
• **Security posture**: One‑time OAuth authentication isolates GTM credentials, and the repository follows standard Go security practices, though a formal security audit is still advisable.
• **Ecosystem fit**: The server aligns with the growing MCP ecosystem, making it a drop‑in for any AI agent that already supports MCP.

Overall, the project is mature enough for a pilot in a production environment, provided you perform a final license and security review and set up monitoring around the authentication token lifecycle.

Use in a workflow

paolobietolini/gtm-mcp-server helps connect AI assistants to real tools and data through a standard protocol.

• connect AI agents to tools
• ship Model Context Protocol servers
• standardize integrations

Limitations and review notes

High for an OSS candidate: recent activity, adoption, and ecosystem signals are strong enough for a serious pilot.