Skip to content
Coframe Agent Help Center

What is editor-web?

Coframe Agent is an AI-powered editor for building, previewing, and shipping changes to a website — without hand-writing every line of code. You describe what you want in plain language, the editor makes the change, and you see it running live in an isolated preview before anything ships.

Who it’s for

Section titled “Who it’s for”
  • Marketers and growth teams who want to test new copy, layouts, and designs quickly — and run A/B tests to prove what works.
  • Designers who want to iterate on a real, running page instead of a static mockup — with visual annotations for precise feedback.
  • Engineers who want an AI agent that plans, delegates, and tests — not just a code autocomplete.
  • Product managers who want to define requirements upfront and track the agent’s progress against them.

What you can do

Section titled “What you can do”
  • Describe a change in chat and have the agent build it for you.
  • Preview it live in a real Chromium browser, exactly as your visitors would see it.
  • Watch the plan — the agent breaks your request into tasks, each mapped to your requirements.
  • Delegate to sub-agents — complex changes are split across specialist agents working in parallel.
  • Test automatically — vitest runs in the editor, with results mapped to your requirements.
  • Annotate visually — click on elements in the preview and leave feedback the agent understands.
  • Extend with skills — install reusable skills that teach the agent your processes and preferences.
  • Iterate — refine the result with follow-up instructions until it’s right.
  • Publish the change when you’re happy with it.

What makes editor-web different

Section titled “What makes editor-web different”

Works on your existing site

Unlike tools that only generate new projects, Coframe Agent edits your production codebase — creating variants you can A/B test.

Real browser preview

A full Chromium instance renders your page — JavaScript, API calls, third-party scripts, and interactions all work.

Sub-agents

The agent delegates independent tasks to specialist sub-agents that work in parallel, with full observability.

Built for teams

Requirements capture, task planning, visual annotations, and testing give non-technical stakeholders real control.

How it fits together

Section titled “How it fits together”

Each piece of work happens inside an edit session. A session is your workspace: it holds the conversation, the code being edited, the task plan, and a live preview of the page. Sessions are self-contained, so you can work on different changes independently.

The workspace is a multi-pane IDE with eight panels — chat, tasks, requirements, preview, files, sub-agents, skills, and tests — all reflecting the same session state in real time.