Devtools logo

Devtools

OrganizationPopular
vercel-labs
devtools

Drop-in inspector panel for any json-render app. Use when the user wants to debug a generative UI, inspect the spec tree, edit state at runtime, see dispatched actions, follow stream patches live, browse a catalog, or pick DOM elements to find their spec keys. Triggers include "add devtools", "debug json-render", "inspect the spec", "why is this element not rendering", "see the state at runtime", or requests to tap streams / capture action logs for `@json-render/devtools`.

Overview

Publishervercel-labs
Repositoryjson-render
Skill namedevtools
Stars
16.5K
Forks
887
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by vercel-labs on GitHub. Read the source before you install it.

Installation

Install the Devtools AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/vercel-labs/json-render.git /tmp/json-render
mkdir -p .claude/skills
cp -r /tmp/json-render/skills/devtools .claude/skills/devtools
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Devtools in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Devtools on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Devtools is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

@json-render/devtools

A floating inspector panel for json-render apps. Framework-agnostic core + per-framework adapters (React, Vue, Svelte, Solid).

Production-safe: the component renders null when NODE_ENV === "production".

Install

Install the core package plus the adapter that matches the host app's renderer.

bash
# React
npm install @json-render/devtools @json-render/devtools-react

# Vue
npm install @json-render/devtools @json-render/devtools-vue

# Svelte
npm install @json-render/devtools @json-render/devtools-svelte

# Solid
npm install @json-render/devtools @json-render/devtools-solid

Drop-in usage

Place <JsonRenderDevtools /> anywhere inside the existing <JSONUIProvider> (or framework equivalent). No other wiring required.

React

tsx
import { JsonRenderDevtools } from "@json-render/devtools-react";

<JSONUIProvider registry={registry} handlers={handlers}>
  <Renderer spec={spec} registry={registry} />
  <JsonRenderDevtools spec={spec} catalog={catalog} messages={messages} />
</JSONUIProvider>;

Vue

vue
<script setup>
import { JsonRenderDevtools } from "@json-render/devtools-vue";
</script>

<template>
  <JSONUIProvider :registry="registry">
    <Renderer :spec="spec" :registry="registry" />
    <JsonRenderDevtools :spec="spec" :catalog="catalog" :messages="messages" />
  </JSONUIProvider>
</template>

Svelte

svelte
<script>
  import { JsonRenderDevtools } from "@json-render/devtools-svelte";
</script>

<JSONUIProvider {registry}>
  <Renderer {spec} {registry} />
  <JsonRenderDevtools {spec} {catalog} {messages} />
</JSONUIProvider>

Solid

tsx
import { JsonRenderDevtools } from "@json-render/devtools-solid";

<JSONUIProvider registry={registry}>
  <Renderer spec={spec()} registry={registry} />
  <JsonRenderDevtools
    spec={spec()}
    catalog={catalog}
    messages={messages()}
  />
</JSONUIProvider>;

Controls

  • Floating toggle appears bottom-right.
  • Hotkey: Ctrl/Cmd + Shift + J (configurable via hotkey prop).
  • Drawer is resizable; height persists to localStorage.

Props

  • spec (Spec | null) — current spec.
  • catalog (Catalog | null) — catalog definition; required for the Catalog panel.
  • messages (UIMessage[]) — AI SDK useChat messages; scanned for spec data parts.
  • initialOpen (boolean) — start open.
  • position ("bottom-right" | "bottom-left" | "right") — dock + toggle corner. "bottom-*" docks at the bottom; "right" docks at the right edge full-height (recommended for app-shells that already use 100vh or fixed bottom bars).
  • hotkey (string | false) — "mod+shift+j" by default.
  • bufferSize (number) — event ring-buffer cap, default 500.
  • reserveSpace (boolean, default true) — when true the panel pushes the host app by applying padding-bottom / padding-right on body. Set to false to keep the panel as a pure overlay.
  • allowDockToggle (boolean, default true) — show a toolbar button so the user can flip the panel between bottom-dock and right-dock. User choice persists to localStorage and overrides position on subsequent mounts. Pass false to lock the dock to position.
  • onEvent ((DevtoolsEvent) => void) — optional tap.

Panels

  • Spec — element tree rooted at spec.root; props/visibility/events/watchers detail; integrated validateSpec warnings.
  • State — every JSON Pointer path with inline edit via store.set.
  • Actions — dispatched actions timeline (name, params, result/error, duration).
  • Stream — spec patches, text chunks, token usage, lifecycle markers grouped by generation.
  • Catalog — components + actions declared in the catalog with prop chips.

Picker (toolbar)

The element picker is a toolbar button in the panel header (Chrome-DevTools-style), not a tab. Click it to activate pick mode, then click any rendered element in the page — selection jumps to the Spec tab with that element focused. Esc cancels.

Reserved space & docking

The panel can dock at the bottom or the right edge, and by default the user can flip between the two with a toolbar button (the choice persists to localStorage). Set allowDockToggle={false} if the host app only works with one dock — the button is hidden and the dock is locked to position.

Pick an initial dock that fits your layout:

  • Bottom dock (default) — works best for docs / marketing / content-flow sites and for app shells built with a height: 100% chain (html { height: 100% }body { height: 100% }.app { height: 100% }). The panel writes its height to --jr-devtools-offset-bottom and applies matching padding-bottom to body, so non-fixed content naturally makes room.
  • Right dock (position="right") — recommended for app-shell layouts that use 100vh or position: fixed; bottom: 0. Right docking sidesteps the bottom edge entirely and writes its width to --jr-devtools-offset-right instead.

Apps that use 100vh, position: fixed, or position: sticky can opt specific elements in with the published CSS custom properties:

css
.composer   { bottom: var(--jr-devtools-offset-bottom, 0); }
.sidebar    { right:  var(--jr-devtools-offset-right,  0); }
.app-shell  { height: calc(100vh - var(--jr-devtools-offset-bottom, 0)); }

If the automatic body padding causes problems with a particular layout, pass reserveSpace={false} to make the panel a pure overlay — the CSS custom properties are still published so you can reserve space manually.

(--jr-devtools-offset is kept as a back-compat alias for whichever edge is currently active.)

Multiple renderers on one page (e.g. a chat)

A single <JsonRenderDevtools /> can inspect many <Renderer /> instances at once — a chat where each assistant message renders its own spec, a dashboard made of several independent widgets, etc. The recipe:

  1. One top-level <JSONUIProvider> so every renderer shares one state store and one action dispatcher. Devtools lives inside this provider and sees everything through it.
  2. Per-renderer specs, shared state — each assistant message renders <Renderer spec={msgSpec} registry={registry} /> directly, not wrapped in its own StateProvider. State paths from different messages must not collide.
  3. Namespace state per turn — when the source is an AI stream, hand the agent a unique messageId and require every element key (<id>-root) and state path (/<id>/count) to be prefixed with it.
  4. Pass spec={latest} + messages={all}spec drives the Spec panel (usually the newest assistant message's spec), while messages feeds the Stream panel with patches from every turn.
  5. Actions and the picker are already globalregisterActionObserver captures dispatches from any ActionProvider in the tree, and data-jr-key is written by the renderer itself, so Pick works across every rendered element regardless of which message produced it.

See examples/devtools for a full AI chat wired this way.

Imperative API (React only)

tsx
import { useJsonRenderDevtools } from "@json-render/devtools-react";

const devtools = useJsonRenderDevtools();
devtools?.open();
devtools?.toggle();
devtools?.recordEvent({ kind: "stream-text", at: Date.now(), text: "hi" });

Returns null in production or before the component mounts.

Server-side stream tap

Capture spec patches at the API route so events persist server-side or flow into your own telemetry.

ts
import { tapJsonRenderStream, createEventStore } from "@json-render/devtools";
import { pipeJsonRender } from "@json-render/core";

const events = createEventStore({ bufferSize: 1000 });
const tapped = tapJsonRenderStream(result.toUIMessageStream(), events);
writer.merge(pipeJsonRender(tapped));

YAML equivalent: tapYamlStream.

Under the hood

  • Shadow-DOM isolated panel — the panel's styles never leak into the host app and vice versa.
  • Ring-buffered event store — capped log of devtools events (state changes, action dispatches, stream patches, etc.).
  • Action observer registry — each framework's ActionProvider reports via notifyActionDispatch / notifyActionSettle in @json-render/core; devtools subscribes via registerActionObserver.
  • Picker element tagging — while devtools is mounted, ElementRenderer wraps each rendered element in <span data-jr-key="..." style="display:contents"> so the picker can map DOM → spec key. No layout impact.

Frequently asked questions

What does the Devtools AI skill do?

Drop-in inspector panel for any json-render app. Use when the user wants to debug a generative UI, inspect the spec tree, edit state at runtime, see dispatched actions, follow stream patches live, browse a catalog, or pick DOM elements to find their spec keys. Triggers include "add devtools", "debug json-render", "inspect the spec", "why is this element not rendering", "see the state at runtime", or requests to tap streams / capture action logs for `@json-render/devtools`.

Why use Devtools on TypingMind?

Because you install it once and use it with any model. Devtools is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Devtools in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/vercel-labs/json-render/tree/main/skills/devtools. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Devtools?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Devtools?

As many as you like. As long as a model supports skills, you can use Devtools with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Devtools AI skill free?

Yes. It is published on GitHub by vercel-labs under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇