Analytics made for Agents.

Add user insight to any codebase with a single prompt. No dashboard. Machine readable. No cookies.

Tell your coding agent ↴

› use luku.io

01 / Installation

Ask your agent to install Luku.io

Paste one prompt into Cursor, Claude Code, or any agent. It creates the site, adds the script tag, and verifies the first event before you deploy.

Agent chat installing Luku

02 | Get data

Human understanding. Made for machines.

Ask what visitors notice, skip, or try to click. The answer arrives as structured data in the same chat where you make changes, never a dashboard you have to read.

Agent answering what is breaking on the site

03 / Run

No dashboards. No hassle. Just key insight agents need improving sites and apps.

Mark a release, ship the change, and compare before and after. Attention, friction, and conversions from real humans, with bots kept apart. “Make it better” becomes a measurable loop.

Before and after comparison of a release

Changelog

Luku.io is in open beta. Please join us.

  1. Analytics reads now follow the schema, not hand-copied slots
  2. Site-scoped tokens for collaborators
  3. Human-readable docs site generated from the agent docs
  4. Install proof you can trust: verify reports every stage
  5. Localhost and private networks never count as visitors
  6. Release comparison as an MCP tool
  7. Bot and AI-crawler classification

Start with a prompt

Add user insight to any codebase with a single prompt.

Open the project you want to measure. Your agent handles the install, verification, and first answer.

Paste into Cursor, Claude Code, or any agent

› use luku.io