Introducing ConsentLayer: consent management built for the AI-first web

Mircea Sandu

Today we are launching ConsentLayer, a consent management platform for teams that ship fast and increasingly build with AI coding agents. It handles the cookie banner, blocks trackers until visitors say yes, scans your site for what is actually loading, applies the right rules by region, and records every consent decision so you can prove what happened.

We built it because the way most teams add consent tooling has stopped matching the way they build.

Consent tooling breaks in the agent era

Adding a cookie banner used to be a one-time chore. You picked a tool, pasted a snippet, and hoped the categories were right. That model already leaked: the snippet went stale, new analytics scripts slipped in unblocked, and nobody re-checked which region saw which banner.

Now developers hand more of that work to AI coding agents. You describe a change in your editor and the agent writes the code. But consent has stayed a manual, dashboard-only task. Your agent can scaffold an app in minutes and then stall on the one thing that carries real regulatory weight.

ConsentLayer closes that gap. Everything you can do in the dashboard, an agent can do from your editor through our MCP server. The consent layer becomes part of the same workflow as the rest of your stack.

What ConsentLayer actually is

Start with the parts you already expect from a consent tool, done properly.

A consent banner you control. Three templates (floating, wide bar, and a modal on paid plans) with seven positions, full color, text, and logo customization, and per-button visibility. After the banner closes, a floating “Manage preferences” button stays available so visitors can change their mind.

Real script and content blocking. ConsentLayer blocks tracking scripts until a visitor consents, with 20 built-in rules covering services like Google Ads, Google Analytics, Facebook, TikTok, Hotjar, Mixpanel, and Plausible, plus custom rules per site. It gates embeds and iframes the same way, and it maps consent to Google Consent Mode v2 signals and honors Global Privacy Control.

A scanner that tells you the truth. Point the cookie scanner at your site and it crawls your pages, detects known services from the built-in library, and drops anything it does not recognize into a review queue with suggested categories. It diffs scan to scan so you see what changed. Scans are manual on Free and scheduled on paid plans.

Rules that follow the visitor. Geo-targeted consent rules resolve the visitor’s country from a CDN header (no raw IP stored client-side) and apply the right approach by region, with built-in templates for GDPR opt-in, CCPA/CPRA opt-out with GPC, and LGPD opt-in. Custom groups are available on paid plans.

A consent log you can hand to a reviewer. Every decision is recorded with its categories, jurisdiction, banner version, user agent, and a salted IP hash. Raw IPs are never stored. Filter by country, decision, or date, and export to CSV.

The banner runtime is vanilla JavaScript with zero dependencies. It loads asynchronously and never blocks rendering. It runs in 10 languages out of the box.

The MCP story

Here is the part we are most excited about. ConsentLayer ships an MCP server, so an AI coding assistant can set up consent management for you.

One command connects it:

npx -y @consentlayer/mcp setup

That registers ConsentLayer with your MCP client. After that, you work in plain language. In Claude Code, Cursor, or any MCP client, a prompt like this is enough:

Add ConsentLayer to this site. Scan it, map the tracking scripts to categories, and give me the install snippet.

The agent calls the real tools behind the scenes: create_site to provision a project, scan_site to crawl and detect services, configure_from_scan to map them to consent categories, and get_setup_snippet to return the code to paste. There are more than 20 tools in total, including search_library, import_service, and list_sites.

You still review the result. But instead of tabbing over to a dashboard, you add consent management from the same place you write everything else. And because it is a real MCP server, it works with any MCP client, not just one editor.

Free versus paid

ConsentLayer is free to start, and the developer surface is not gated.

Free gives you one site, 1,000 pageviews a month, and the full API, SDK, and MCP server. Every plan includes those. You can build the whole integration on Free.

Paid plans raise the limits and unlock a few extras. Solo ($19/mo) and up add the modal banner template, the option to remove ConsentLayer branding, scheduled scans, and custom geolocation groups. Studio ($79/mo) adds more sites, pageviews, and team members. Agency adds hosted cookie policy pages for managing many client sites at once. Annual billing saves 20% across the board.

There is no separate developer tier to buy. The API, SDK, and MCP server are there from the first minute on Free.

Get started

Create a free account and connect the MCP server, or paste a single script tag if you prefer to do it by hand. Either path takes a few minutes.

If you have moved off another consent platform and want a second pair of eyes on your setup, run the scanner and compare the results. We think you will like what you find.