Open Brain
GitHubGitHub RepoNeeds Review

pbakaus/impeccable

Impeccable gives coding agents a shared design vocabulary, project context, and frontend checks. Builders can use it to make design reviews and polish passes more repeatable across supported tools.

GitHub
Open Source Open MarkdownOpen JSON
Source Summary

Impeccable provides **1 skill and 24 commands** for design guidance, plus **62 deterministic detector rules** for AI-generated frontend design. Its setup records durable product context in PRODUCT.md and keeps visual direction separately in DESIGN.md.

Practical Implication

Builders can use commands such as audit, critique, and polish to guide reviews, and run the deterministic checks through the CLI or browser extension without an API key. The project supports several coding tools, including Claude Code, Cursor, and Codex.

Agent-Ready Context
Impeccable provides **1 skill and 24 commands** for design guidance, plus **62 deterministic detector rules** for AI-generated frontend design. Its setup records durable product context in PRODUCT.md and keeps visual direction separately in DESIGN.md.

Builders can use commands such as audit, critique, and polish to guide reviews, and run the deterministic checks through the CLI or browser extension without an API key. The project supports several coding tools, including Claude Code, Cursor, and Codex.

The rules are deterministic, while some critique checks use an LLM. Setup and hook behavior vary by tool; for example, Codex users may need to approve the project hook after installation or updates.
Connected Context · Feed7 Judgment

Impeccable combines durable product and visual context with design commands and deterministic frontend checks, with some critique checks using an LLM. Its tool-specific setup and hook approval requirements mean portability still involves integration work; the supplied details describe capabilities, not measured improvements in design quality.

Context Map
agentcoding#skills#design-engineering#coding-agents
Uncertainty
The rules are deterministic, while some critique checks use an LLM. Setup and hook behavior vary by tool; for example, Codex users may need to approve the project hook after installation or updates.