App Builder

Community

Undertone0809

app-builder

Create and iteratively improve local web products from natural-language requests, then prepare them to run as Rudder Apps. Use Rudder's maintained scaffold for new websites and preserve the existing stack for Next.js, React, Vue, Astro, SvelteKit, Nuxt, and other local web projects.

Overview

Publisher	Undertone0809
Repository	`rudder`
Skill name	`app-builder`
Stars	290
Forks	31
Bundled files	59
License	Apache-2.0
Links	Skill source Repository Homepage

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.
59 bundled files
Scripts, templates, and references the model can read while it works. Files are read-only and never executed.
Open source
Published by Undertone0809 on GitHub. Read the source before you install it.

Installation

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

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.

Open the app and go to Plugins → Skills.
Choose "Install from GitHub".
Paste the skill folder URL below and confirm.
Enable the skill in any chat where you want it available.

Store

Installed

Skills

Add skills

Import one or more skills from a public GitHub folder or a local .skill or .zip archive, or SKILL.md file.

From GitHub URLFrom .skill, .zip, or SKILL.md

GitHub repo or folder

https://github.com/Undertone0809/rudder/tree/main/server/resources/bundled-skills/app-builder

Points to a single skill folder, or any folder that contains multiple skills. Public repos only.

CancelContinue

Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.

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/Undertone0809/rudder.git /tmp/rudder
mkdir -p .claude/skills
cp -r /tmp/rudder/server/resources/bundled-skills/app-builder .claude/skills/app-builder

Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable App Builder 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 App Builder 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 App Builder 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.

App Builder

Create or improve a useful local web product without making a non-technical user choose a framework, database, package manager, or process topology.

A Rudder App is an ordinary local website presented as an App inside Rudder. This Skill owns requirements, source changes, and product verification. Rudder Desktop owns folder selection, launch-definition review, process lifecycle, loopback attestation, and embedded opening.

Build Workflow

Capture a compact brief in the conversation: user/job, primary workflow, essential screens, durable data, external integrations, constraints, and exclusions. Ask at most three grouped questions, and only when an answer materially changes behavior, real-data safety, or external authority.
Choose the source path:
- For a new App, use the maintained scaffold in assets/scaffold/. Prefer the Apps workspace flow when it is available. Otherwise run scripts/scaffold.mjs with an explicit workspace root and target. Never recreate the foundation from memory or overwrite a non-empty directory.
- For an existing web project, inspect its current framework, package manager, development script, data boundary, and tests before editing. Preserve its stack and conventions; never replace it with the maintained scaffold merely to make it a Rudder App. When the originating prompt includes Rudder App handoff, retain its App ID for the completion protocol below. Do not guess an App ID or use one from a different conversation.
Read references/scaffold-contract.md before changing the runtime, database foundation, manifest, health endpoint, import/export contract, or background-job runner.
Implement the user's domain model and workflows. Keep the generated project conventional and independently runnable.
Follow references/data-safety.md whenever an app already has data, imports user data, changes its schema, or needs a real-data diagnosis.
Follow references/design-guidelines.md for visible UI work. For a new App, also read references/rudder-ui-preset.md and start from the versioned Rudder UI preset already present in the maintained scaffold. Keep its semantic tokens and source-owned shadcn primitives unless the brief explicitly requires a different product identity. For an existing project, preserve its design system by default; align it to Rudder only when the user asks for that change. Prefer coherent workflow screens over a generic dashboard assembled from decorative cards.
Run migrations against development or snapshot data, then run typecheck, unit tests, build, and relevant app E2E tests. Review the final source and rendered product for correctness, regressions, usability, data safety, and evidence integrity; resolve every blocking finding before handoff.
Prepare the website for its Rudder Apps handoff:
- A new managed App created from Apps Home keeps its typed manifest. After the completion gate passes, report verified_source_ready with the originating App ID so Rudder Desktop can verify, start, and open that App automatically. Do not load the same source through the manual Local App path.
- An independently authored existing project keeps a direct supported development script. Add the package.json rudder readiness or open-path fields only when discovery cannot infer them safely, then direct the operator to Apps + > Add local web project to select the folder and review the launch definition. The Agent writes and verifies source; Desktop alone owns runtime admission and process start. Do not invent a public URL, tunnel, cloud deployment, or unreviewed launch command.
Read references/verification.md, then verify the rendered app with Rudder Browser. Exercise the main workflow plus at least one production-shaped edge case and inspect console errors.
Materialize final screenshots and validation evidence in the originating Chat or Run. A localhost URL or tool-only screenshot is not durable output.
For an originating managed App handoff, use the injected run context without printing or requesting credentials:
- Run node scripts/report-build-status.mjs building <app-id> when implementation begins.
- Run node scripts/report-build-status.mjs verified_source_ready <app-id> only after every Completion Gate item passes. This is the machine-visible signal that authorizes Desktop to verify, start, and open the exact assigned source root.
- Run node scripts/report-build-status.mjs failed <app-id> when work ends in a terminal failure. Do not emit verified_source_ready for a question, partial implementation, canceled run, missing evidence, or blocker. Finish by telling the user the App is opening and that they can continue improving it in this Chat. If the status command fails, report that failure honestly instead of claiming the App opened.

Real Data Decision

New app with no user data: use the scaffold's development database and synthetic fixtures.
UI or ordinary logic change on an app with user data: verify against a temporary database snapshot by default.
Diagnosis that genuinely depends on user records: ask whether to use the original database, a snapshot, or a redacted copy. State that relevant data may be sent to the configured model provider.
Schema change: snapshot first and rehearse the migration on the snapshot. Applying it to user data is an application-specific action and requires explicit user intent.
A direct operation on the formal database requires explicit user intent.

Never claim the Agent is sandboxed from files it can access. Use product controls and least-data workflow choices; explain the boundary honestly.

Runtime Boundary

App source, data, builds, and execution stay on the user's device.
The app may use outbound APIs only when requested and configured.
Background tasks implemented by the App run only with that managed App process. V1 does not provide a separate Rudder job scheduler or catch-up UI.
Never place secret values in source, rudder.app.json, tool arguments, Chat, screenshots, test fixtures, or logs. V1 does not provide a Rudder Secret binding UI, so ask the user how the App should obtain any required integration credential.
Opening Chat, Apps Home, or a saved view must not passively start an app. A run-scoped verified_source_ready handoff may start and open its managed App.
Cloud builds, hosted runtimes, public preview links, custom domains, managed cloud databases, and cross-device synchronization are outside this skill.

Completion Gate

Do not call the app complete until:

source and any required manifest or package.json Rudder configuration are durable;
typecheck, unit tests, and build pass;
blocking code and product review findings are resolved;
the reviewed readiness path is ready on an attested loopback process;
Browser verification covers the primary workflow and a relevant edge case;
the versioned Rudder UI preset check passes for a new maintained App;
persistence behavior is verified against development or snapshot data;
current screenshot evidence is attached to the work;
user data was not modified without explicit intent; and
stop/cleanup leaves no owned process behind.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the App Builder AI skill do?

Why use App Builder on TypingMind?

Because you install it once and use it with any model. App Builder 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 App Builder in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Undertone0809/rudder/tree/main/server/resources/bundled-skills/app-builder. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use App Builder?

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 App Builder?

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

Is the App Builder AI skill free?

Yes. It is published on GitHub by Undertone0809 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

Next Cache Components Adoption

vercel

OrganizationPopular

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the `cacheComponents` flag, work…

Next Cache Components Optimizer

vercel

OrganizationPopular

Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal…

Next Partial Prefetching Adoption

vercel

OrganizationPopular

Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the `partialPrefetching` flag, opt routes…

Next Partial Prefetching Optimizer

vercel

OrganizationPopular

Optimize what selected Next.js client navigations include before the click under Partial Prefetching. Use after Cache Components and Partial Prefetching are adopted when the user wants selected…

Overview

Installation

Install in TypingMind

Install in another agent

Use it in TypingMind

Works with any AI model

Skill instructions

App Builder

Build Workflow

Real Data Decision

Runtime Boundary

Completion Gate

Bundled files

Frequently asked questions

Related AI Skills

Set up your own AI workspace now