This product was not featured by Product Hunt yet. It will not be visible on their landing page and won't be ranked (cannot win product of the day regardless of upvotes).
Product upvotes vs the next 3
Waiting for data. Loading
Product comments vs the next 3
Waiting for data. Loading
Product upvote speed vs the next 3
Waiting for data. Loading
Product upvotes and comments
Waiting for data. Loading
Product vs the next 3
Loading
AI-native design engine · Reframe
Programmable design engine for AI agents. Open-source.
AI-native. Open-source. MCP-native. Like Claude Design. In-canvas agent spawns from your Claude Code subscription — no API keys, no token bills. Chat UI + right-click macros: rebrand, theme, scale, rotate, fix, etc. Your AI writes HTML + DESIGN.md → typed INode AST → 37-rule audit + 8 aesthetic scores → 8 export formats. Deterministic. 6 MCP tools: design, compile, inspect, edit, export, project. Claude Code, Cursor, any MCP agent. AGPL-3.0. For devs who want design in git.
I've been building this for months. Strange coincidence — Anthropic shipped Claude Design the same day I pushed Reframe v0.1. Trying their product made something clear: we're at different layers. Claude Design is a product: polished, SaaS, consumer-facing.
Reframe is an engine: typed AST, deterministic, open-source. Different audiences, different goals.
The gap that got me building:
Every AI design tool today works on HTML strings. Prompt → generate → edit → regenerate. After 10 iterations you've drifted: colors shifted, spacing off-grid, brand tokens leaked. No way to measure it, no way to prevent it.
Reframe keeps every scene in a typed INode AST. Every edit is a deterministic transform — not an LLM re-generation. Every export is byte-equal across runs. Audit findings are numbers, not vibes.
What works today (v0.1):
Runs on YOUR Claude Code subscription — no API keys, no token bills
→ Interactive CanvasKit viewport — Figma-like select, drag, resize, text editing → 37-rule quality audit + 8 aesthetic metrics + brand fidelity scoring → 8 export formats: HTML, React, SVG, PNG, PDF, Lottie, animated HTML, site bundle → 6 MCP tools for Claude Code, Cursor, and any MCP-compatible agent → 60+ brand design systems included (Stripe, Linear, Airbnb, Apple…)
What's rough (v0.1 is v0.1):
→ Tailwind/styled-components export: scaffolded, not full utility-class emission yet
→ Platform UI: functional but spartan — engine is the mature part
→ No hosted version — self-host, clone the repo
→ Active bugfix mode — file issues, I patch live during launch week
Built alone on nights and weekends.
Who this is for: • Teams with brand guidelines that LLM-drift keeps breaking • Devs building their own AI-design tools internally • Regulated industries that need audit trails on visual assets • Anyone who wants design work in git
Why AGPL-3.0: fork freely, self-host, modify. But if you run it as a SaaS you open-source your layer. Prevents commercial clone-and-run. Keeps the ecosystem honest.
Looking for: ★ GitHub stars if the approach resonates 💬 Feedback on what's broken / what's missing 🤝 Contributors — MCP integrations, new export formats, language bindings
Claude Design is brilliant for ideation. Reframe is what you pipe it INTO when you need to ship.
If "AI design as a typed engine" resonates — whether you want design in git, embed design into your product, or build agents that operate on typed scenes — I'd love feedback. Issues / discussions open in the repo.
About AI-native design engine · Reframe on Product Hunt
“Programmable design engine for AI agents. Open-source.”
AI-native design engine · Reframe was submitted on Product Hunt and earned 1 upvotes and 1 comments, placing #64 on the daily leaderboard. AI-native. Open-source. MCP-native. Like Claude Design. In-canvas agent spawns from your Claude Code subscription — no API keys, no token bills. Chat UI + right-click macros: rebrand, theme, scale, rotate, fix, etc. Your AI writes HTML + DESIGN.md → typed INode AST → 37-rule audit + 8 aesthetic scores → 8 export formats. Deterministic. 6 MCP tools: design, compile, inspect, edit, export, project. Claude Code, Cursor, any MCP agent. AGPL-3.0. For devs who want design in git.
On the analytics side, AI-native design engine · Reframe competes within Design Tools, Developer Tools and Artificial Intelligence — topics that collectively have 1.2M followers on Product Hunt. The dashboard above tracks how AI-native design engine · Reframe performed against the three products that launched closest to it on the same day.
Who hunted AI-native design engine · Reframe?
AI-native design engine · Reframe was hunted by Ilya Makarov. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
For a complete overview of AI-native design engine · Reframe including community comment highlights and product details, visit the product overview.
Ilya Makarov
Hey PH 👋 I'm Ilya, solo dev behind Reframe.
I've been building this for months. Strange coincidence — Anthropic shipped Claude Design the same day I pushed Reframe v0.1. Trying their product made something clear: we're at different layers. Claude Design is a product: polished, SaaS, consumer-facing.
Reframe is an engine: typed AST, deterministic, open-source. Different audiences, different goals.
The gap that got me building:
Every AI design tool today works on HTML strings. Prompt → generate → edit → regenerate. After 10 iterations you've drifted: colors shifted, spacing off-grid, brand tokens leaked. No way to measure it, no way to prevent it.
Reframe keeps every scene in a typed INode AST. Every edit is a deterministic transform — not an LLM re-generation. Every export is byte-equal across runs. Audit findings are numbers, not vibes.
What works today (v0.1):
→ Interactive CanvasKit viewport — Figma-like select, drag, resize, text editing
→ 37-rule quality audit + 8 aesthetic metrics + brand fidelity scoring
→ 8 export formats: HTML, React, SVG, PNG, PDF, Lottie, animated HTML, site bundle
→ 6 MCP tools for Claude Code, Cursor, and any MCP-compatible agent
→ 60+ brand design systems included (Stripe, Linear, Airbnb, Apple…)
Who this is for:
• Teams with brand guidelines that LLM-drift keeps breaking
• Devs building their own AI-design tools internally
• Regulated industries that need audit trails on visual assets
• Anyone who wants design work in git
Why AGPL-3.0: fork freely, self-host, modify. But if you run it as a SaaS you open-source your layer. Prevents commercial clone-and-run. Keeps the ecosystem honest.
Looking for:
★ GitHub stars if the approach resonates
💬 Feedback on what's broken / what's missing
🤝 Contributors — MCP integrations, new export formats, language bindings
Claude Design is brilliant for ideation. Reframe is what you pipe it INTO when you need to ship.
If "AI design as a typed engine" resonates — whether you want design in git, embed design into your product, or build agents that operate on typed scenes — I'd love feedback. Issues / discussions open in the repo.
github.com/ilya-makarov-dev/Reframe
Thanks for checking it out 🙏