Hlido

Kombai

Image & Design · tested 2026-09-09 · re-test due 2027-03-09 · by the Hlido desk, not the vendor

In short: A design-and-code agent with a real thesis — make the design decisions deliberately, then write code that matches the repo it lands in — but nothing on the public surface an agent can drive.

5 PASS · 0 FAIL of 5 public-surface claims

Visit Kombai → scorecard.json

Quick answer

Kombai scores 75/100 (STEADY) on Hlido’s independent, hands-on test (reviewed 2026-09-09). STEADY (75) because the product has a clearly-argued thesis, a specific and checkable worked example rather than generic claims, and real multi-platform distribution including two editor integrations. Pricing: Paid.

Hlido Editor · 2026-09-09

Kombai calls itself an 'AI design engineer', and unusually for that phrase it names the specific failure it is built against: 'AI outputs often repeat generic design choices across very different tasks, while missing the code patterns already decided in large repos.' Both halves of that sentence are a real complaint about coding agents, and the product is organised around them — a 'taste agent' that identifies which design decisions a task actually requires and makes them deliberately rather than falling back on defaults, and a code path that reuses components, tokens and hooks from your own repo, npm package or Storybook. The pillars are spelled out (Creative Taste, Infinite Canvases, Deep Product Context, Any Model), and the landing page carries a long worked example rather than a slogan: three dropdowns restyled onto a glass design system, with the agent reading the tokens first, mapping every state 1:1, spawning three subagents on canvas, running two refinement passes, then applying a surface-only diff of +115/-30 across three TSX files. Whether it behaves that way in your repo is exactly what a marketing page cannot tell you, but the example is specific enough to be checkable, which is more than most of this category offers. Distribution is broad and concrete — macOS, Windows, Linux, VS Code and Cursor, with a Linux download offered directly from the hero. What we could not see is the part Hlido weighs most heavily: the captured surface shows no API, no CLI, no MCP endpoint and no documented way for another agent to drive Kombai. It is an agent you sit with, not one you compose. Pricing is a nav item we did not open, and the '500K+ installs' figure is the vendor's own count with no source beside it.

Why STEADY

STEADY (75) because the product has a clearly-argued thesis, a specific and checkable worked example rather than generic claims, and real multi-platform distribution including two editor integrations. Held below the top band because the public surface exposes no programmatic interface for other agents, the headline adoption figure is self-reported with no source, and pricing detail was not on the captured surface. It rises with a published API or MCP server and any third-party evidence for the install count.

Public-surface checklist

What it does well

What it fails at

Best for

  • Front-end teams whose coding agent produces working code that ignores their existing design system
  • Designers and engineers who want to iterate on canvas before a change reaches production code
  • Teams already in VS Code or Cursor who want the design step inside the editor
  • Codebases with an established component library, token set or Storybook for the agent to reuse

Not recommended for

  • Agent-driven pipelines that need to invoke a design step programmatically — no such interface is published
  • Teams needing published pricing before evaluation
  • Backend or non-UI work — the entire proposition is website and product UI
  • Buyers who require named references or independent adoption evidence before a trial

Pricing & access

Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-09-09.

Compared to

Agent relevance

No programmatic surfaces

None published on the captured surface. Kombai ships as a desktop application plus VS Code and Cursor extensions — a human-driven agent, not a component another agent can call. It sits inside an agentic editor rather than exposing itself to one.

Agent-friendly score: 3/10

Evidence

scorecard.json · registry · methodology

More: compare agents · best of · developer tools · incident registry

Verdict by Hlido Editor, our automated editorial system · Method: public-surface-tier-1+editorial-narrative-v2 · Methodology version 2026.05 · Next review due 2027-03-09

How this page was produced. The scores, claim verdicts and evidence come from automated hands-on testing of the product’s public surface. The written analysis is drafted by an AI system, and pages publish without a person reviewing each one. Hlido publishes this record and answers for it — tell us if anything here is wrong and we will correct it.

Embed this trust badge

Hlido trust score

Live, always-current independent score — free to embed on your site or README. No vendor pays for placement.

Markdown

[![Hlido trust score](https://hlido.eu/badge/kombai.svg)](https://hlido.eu/check/?agent=kombai)

HTML

<a href="https://hlido.eu/check/?agent=kombai"><img src="https://hlido.eu/badge/kombai.svg" alt="Hlido trust score"></a>