Project overview
Impeccable is a frontend design guidance project for AI coding agents. It provides an installable skill, design commands, and a standalone detector rather than a resource directory. It helps coding agents understand product context, plan interfaces, and inspect and refine existing designs.
Its commands support design reviews, layout changes, copy improvements, and visual iteration in the browser. Deterministic checks in the command-line tool and browser extension run without a large language model or an API key; this does not mean every AI design operation works without a model. The repository lists the Apache-2.0 license. The supplied material does not describe video generation.
Use cases
Impeccable is intended for building or maintaining websites, landing pages, blogs, settings pages, and checkout forms with an AI coding agent. At the start of a project, it can organize audience and product goals. During interface changes, it can review visual hierarchy, layout, copy, and edge cases. Before delivery, it can help refine the design and run quality checks.
Live browser iteration works with a local project through a development server or local static HTML. The README does not support attaching this mode to a deployed production website. Production pages can be inspected with the detector or browser extension, but that inspection does not provide live variant editing or write changes back to source code.
What it does and problems it solves
Impeccable addresses recurring template-like design patterns and quality issues in AI-generated interfaces, such as repetitive color choices, excessive nested cards, cramped spacing, long text lines, small touch targets, and skipped heading levels.
The initialization command records durable product information, including audience, purpose, operating context, and constraints, in PRODUCT.md. Visual systems are recorded separately in DESIGN.md, helping distinguish product facts from visual direction. Design commands cover planning, reviews, typography, layout, motion, error handling, and performance improvements. Detector findings provide evidence for review, but they do not replace inspecting the rendered page or guarantee visual or accessibility quality.
New design flows request visual direction options from an external service. That request still occurs when usage recording is disabled. The README provides a way to disable recording.
How to use
1. After installation, run `/impeccable init` inside your AI coding tool. It inspects the project, asks for missing material product information, and writes PRODUCT.md.
2. Enter `/impeccable` to see the command list. Choose a command for your task, such as `/impeccable shape` to plan an interface, `/impeccable critique landing` to review a landing page, or `/impeccable polish settings` to refine a settings page.
3. Add the area you want to work on after the command. For example, use `/impeccable audit the header` to inspect the header or `/impeccable polish the checkout form` to refine a checkout form. Codex users can view the skill through `/skills` or invoke it with `$impeccable`.
4. To inspect code separately, run `npx impeccable detect src/` in a terminal in your project folder to scan the src/ directory, or run `npx impeccable detect index.html` to inspect a specific HTML file. Use `npx impeccable detect --list-rules` to view the rules, then assess findings alongside the rendered page.
5. For browser-based iteration, use `/impeccable live` only in a local project you trust. Applying copy edits may execute the project's `impeccable:manual-edit-validate` script from package.json. Review that script before using the feature in an unfamiliar project.
6. To disable usage recording, run `npx impeccable telemetry off` in a terminal. Run `npx impeccable telemetry status` to check the current setting. Disabling recording does not stop requests for visual direction options.
Installation
1. Prepare the AI coding tool you want to use and open the root folder of your target project, meaning its outermost project folder. The README lists tools including Cursor, Claude Code, Codex CLI, and GitHub Copilot. Consult the original repository below this page for tool-specific requirements.
2. The recommended installer requires Node and npx to be available. The README does not provide Node installation steps or a minimum version. If these prerequisites are missing, consult the relevant instructions in the original repository first. The skill engine is a self-contained program that needs no separate runtime; it may download the engine on first use.
3. Open a terminal in the project root and run `npx impeccable install`. Follow the prompts to keep or customize the detected AI tools, then choose installation for the current project or user-wide installation for use across projects. The installer also explains automatic design checks and asks whether to install their configuration.
4. Reload your AI coding tool after installation. Codex users must open `/hooks` and approve the project's automatic check configuration when prompted. Grok Build users must complete project-folder trust setup. Then run `/impeccable init` as described in the usage steps.
5. To refresh an existing installation, run `npx impeccable update` and reload the tool. Codex may require approval again if the automatic check configuration changes.
6. If you do not want to use npx, obtain the corresponding download entry from the original repository below this page, or consult its tool-specific manual copying instructions. The README provides these alternatives but does not specify universal operating system requirements or complete source-build steps. For source builds, consult DEVELOP.md in the repository.
Source and usage notes
- Original repository: https://github.com/pbakaus/impeccable
- Source revision: 4281ac224087af05d4d9f0ab20c16a62a345493e
This listing links to the upstream repository. It does not provide a verified download package, and the complete workflow has not been independently validated.
This description was organized and translated with AI from public upstream material. Refer to the original repository for capabilities, setup requirements, costs, and license terms. 
