shadscan

A pre-deployment checker for existing React apps built with shadcn, a component library for React. It audits accessibility, user interactions, loading and error states, forms, responsive layout, and other UI basics.

In plain words
What is it for?
Running a read-only UI audit, reviewing its report, and applying fixes for real findings. It uses the shadscan command-line tool and routes issues to the relevant development skill.
Why use it?
It finds interface problems that may be missed during normal development before users encounter them.

Skill for Claude CodeCodex

Part of the dev-flow plugin — 45 skills shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/lukedj78/dev-flow/shadscan
Any agent
npx skills add lukedj78/dev-flow --skill shadscan
Clone the repo
git clone --depth 1 https://github.com/lukedj78/dev-flow

Made for: Claude Code, Codex.

Or install dev-flow, the plugin that ships this one along with the rest of its 45 skills.

Per session 215 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 5,288 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00215 $0.05288
Opus 5 $0.00108 $0.02644
Sonnet 5 $0.00043 $0.01058
Haiku 4.5 $0.00021 $0.00529

Measured 3d ago against content hash 05b36ac164c3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

shadscan scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

shadscan/SKILL.md · 288 lines

How it starts

The opening of the file, as written. The whole thing — 288 lines — stays where its author put it; the contents beside it link to each section on GitHub.

shadscan — UI-quality & accessibility pre-deploy gate for React/shadcn apps

Runs on a web project that already exists and uses shadcn. It wraps the third-party shadscan CLI — a deterministic static audit of UI fundamentals — and turns its report into applied fixes, routing each finding to the skill that owns it.

This is the gate that covers what neither sibling can see. compliance-audit reads the legal surface, vercel-doctor reads the cost surface; nothing in dev-flow mechanically verified that the UI we prescribe actually got built — that the reduced-motion guard transitions mandates is really there, that the form errors forms specifies are really rendered, that a route has a loading boundary at all.

Third-party tool. shadscan is not an official shadcn product — it's an independent open-source project by TheOrcDev, published to npm as @shadscan/cli (bin shadscan, MIT). Verified against the npm registry + repo README (at time of writing: 0.9.0, ruleset 2026.07.41, report schemaVersion 9 — the report shape is versioned, re-check it). From the README, verbatim: "The default scan is deterministic and read-only. It does not start the app, edit files, call an AI model, upload source, or require application secrets." That makes it safe to run on a private codebase without the --offline dance vercel-doctor needs.

Verified invocation + flags

npx --yes @shadscan/cli                                       # explore: latest, human output
npx --yes @shadscan/[email protected] --json --no-interactive \
  > docs/ui/shadscan.json                                     # audit: PINNED — the mode this skill uses
npx --yes @shadscan/[email protected] --fail-under 70 --no-interactive # CI gate: pinned + ratcheted

⚠️ Pin the version for anything you will compare. npx --yes @shadscan/cli resolves to latest on every call — during a single session it went 0.9.0 → 0.10.0 (ruleset 2026.07.412026.07.42, 59 rules → 60), silently making a before/after diff a comparison across two different rulesets. shadscan knows this: the verification.shadscanCommand it emits in its own acceptance criteria is already pinned (pnpm dlx @shadscan/[email protected] --json). Follow it. Pin the baseline run, pin the re-run, and only drop the pin when you deliberately want the newer ruleset — then re-baseline instead of diffing.

Read the full file on GitHub · 288 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 288 lines · 215 tokens per session scan A 05b36ac164c3

Subscribe to this mod's changes

shadscan is a skill published in the GitHub repository lukedj78/dev-flow (5 stars, last pushed 5d ago), licensed MIT. It adds 215 tokens to every session and 5,288 once invoked, about $0.0011 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

moai-ref-react-patterns

React/Next.js component design patterns, state management strategies, and project structure reference for frontend development. Agent-extending skill that amplifies frontend domain work (spawned via Agent(general-purpose) with frontend instructions) with production-grade React patterns. NOT for: backend API design…

modu-ai/moai-adk · 72 tokens

moai-domain-frontend

Frontend development specialist covering React 19, Next.js 16, Vue 3.5, and modern UI/UX patterns with component architecture. Use when building web UIs, implementing components, optimizing frontend performance, or integrating state management.

modu-ai/moai-adk · 54 tokens

formkit

Use when working with FormKit forms, validation, schema, or custom inputs in React, Vue, or Nuxt projects.

formkit/formkit · 28 tokens

writing-react-native-storybook-stories

Create and edit React Native Storybook stories using Component Story Format (CSF). Use when writing .stories.tsx files, adding stories to React Native components, configuring Storybook addons (controls, actions, backgrounds, notes), setting up argTypes, decorators, parameters, or working with portable stories for…

storybookjs/react-native · 83 tokens

setup-react-native-storybook

Set up Storybook for React Native in Expo, React Native CLI, or Re.Pack projects. Use when adding Storybook to a project, configuring metro.config.js with withStorybook, creating .rnstorybook configuration files, setting up Storybook routes in Expo Router, configuring getStorybookUI, or adding the StorybookPlugin to a…

storybookjs/react-native · 102 tokens

codegen-html

Scaffold and iterate on standalone Preact + HTM applications with zero build dependencies.

initializ/forge · 20 tokens