neuron-explorer
Steering-first neuron explorer for Qwen3.5-4B: push every MLP neuron, keep every rollout, browse the results
Explore the UX. Copy the code. Bring your own idea. Free examples of custom chat experiences for your website, with Stand Chat as the backend.
Developer toolsJavaScript · HTMLGitHub @standchatOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
No updates detected yet. GitHub releases will appear here.
Owners of public repositories under a personal GitHub account can sign in to verify ownership and manage the introduction themselves. No administrator approval is needed.
Sign in, generate a verification file, and commit it to this repository’s default branch within 30 minutes. Files on other branches or forks do not count. Then check the file below to manage the project without administrator approval. Keep the file to retain your access.
If needed, you will go to sign-in. After signing in, return here and press your chosen verification button again.
Run /nomorevibe verify in your project folder to verify ownership and manage the introduction yourself.
Verify in one minuteAsk about other verification options
Examples on how to use Stand Chat
Translation pending
Taken from the repository README
Loading the full README. The saved excerpt is shown below.
Working chat front ends in plain HTML, CSS and JavaScript: animated characters, a CRT terminal, inline components, and a particle renderer.
Live demos · Run the same examples locally: npm start.
Each directory contains a runnable example and an implementation walkthrough. No framework or build step. These were built with AI coding agents; several include the original prompts and build notes.
The UI code is public domain. Conversations use Stand Chat's hosted API, with a shared demo configuration included.Read the full README ↗Working chat front ends in plain HTML, CSS and JavaScript: animated characters, a CRT terminal, inline components, and a particle renderer.
Live demos · Run the same examples locally: npm start.
Each directory contains a runnable example and an implementation walkthrough. No framework or build step. These were built with AI coding agents; several include the original prompts and build notes.
The UI code is public domain. Conversations use Stand Chat's hosted API, with a shared demo configuration included.
npm start
This builds the example list and serves the site at http://localhost:3000. It needs Node 18 or newer and installs nothing.
The examples load Stand with data-stand-id="demo", Stand Chat's shared demo site. It answers on any domain, including localhost, through Stand Chat's demo Stand-in. The status pill in each example's top bar shows what Stand sees. Add ?preview to a URL to show hidden Stand elements while you work on layout.
npx degit standchat/examples/stand-card my-stand-card
Then replace its Stand <script> tag with the installation snippet from Sites in Stand, which carries your own Site ID.
index.html is the front page. Its gallery lists examples.json, which build.mjs generates from each example's <head>. Short browsing descriptions in index.html highlight the reusable chat UX; new contributions use their own metadata automatically. The ORDER list in build.mjs puts the most interesting examples first; any others follow alphabetically. The introduction uses the examples' social images in a collage inspired by _shared/og.png._shared/ holds the example pages' top bar and "How it works" styles. The examples work without it._shared/analytics.js sends visits to stand.chat's PostHog project, on examples.stand.chat only: never on localhost, previews or copied examples. There's no cookie banner: PostHog uses cookies only where stand.chat would (the visitor accepted there, or stand.chat wouldn't ask), and counts everyone else without cookies. The front page and example.js load it._template/ is the starting point for a new example. See CONTRIBUTING.md.og.png) are committed. The build only checks that they are 1200×630; it never regenerates them. To recapture with Chrome, run npm run og for all of them or npm run og -- <folder> for one (og.mjs).node build.mjs and serves the repository as it is (vercel.json).Raw Visitor API clients send POST /v1/events/activate when the visitor opens a custom chat invitation. An inline composer can instead activate on its first submitted message. Discovery, page mount, restoring a conversation, and later messages are not activations. Carry the latest activation ID into POST /v1/sessions; telemetry remains best effort and never blocks chat.
The standalone clients expose activate() and closeInvitation() for actual open/close boundaries, with a first-submit fallback for inline reuse. Keep these UI hooks when copying an example. The character clients retry failed delivery once with the same ID; the inline client family uses its existing best-effort beacon. Copied integrations need the updated client and UI hooks; updating this repository cannot change copies already installed elsewhere. See the custom UI guide for the complete contract.
Run npm test for client accounting and example-model regression checks, and npm run build to validate the catalog.
Public domain under the Unlicense. Copy anything.
Stand Chat's name, logo and mascot aren't covered by it.