One openaiApiKey under runtimeConfig.public, filled at run time from NUXT_PUBLIC_OPENAI_API_KEY, sends your key to every visitor in the page payload. One v-html="route.query.q" on a search page lets anyone send a link that runs scripts as your site. This skill finds those and eight other Nuxt and Vue mistakes before you deploy, and tells you how to fix each one.

Your agent runs a scanner on nuxt.config, app.config, your .env files (including .env.example), every .vue component and page, and the composables, plugins, route middleware, stores and server routes beside them. It knows which runtimeConfig keys are public, follows values from the URL (useRoute, $route, to.query in middleware, getQuery on the server) and from useFetch, useAsyncData and $fetch through the variables that hold them and through v-for aliases, skips comments, and accepts the usual fixes written in code (DOMPurify.sanitize, httpOnly: true, a startsWith('/') check that also rejects // and backslashes, an origin or allow-list check). Public keys such as apiBase, Firebase and VueFire config, Stripe publishable keys, Mapbox pk. tokens and Algolia search keys are left alone, and the default Nuxt starter project passes clean. It never runs Nuxt or your app. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach. Secrets are never printed.

What it checks

BLOCK (fix before shipping)

• N01: secrets sent to the browser (secret-named or secret-valued runtimeConfig.public keys, NUXT_PUBLIC_ variables, app.config, secret keys in components)

• N02: literal secrets in nuxt.config (runtimeConfig or module options, database URLs with a password)

• N03: v-html or innerHTML with route query, params or hash data and no sanitizer

• N04: redirects to URL values (navigateTo with external: true, window.location, window.open, sendRedirect with getQuery) without a check that rejects // and other sites

REVIEW (check before release)

• N11: client source maps (sourcemap true, sourcemap.client true, or vite.build.sourcemap while client maps are on)

• N12: CORS that allows credentials from any origin (route rules, handleCors options, an Origin header copied back)

• N13: plain http:// API calls to public hosts ($fetch, useFetch, baseURL, apiBase, routeRules proxy)

• N14: tokens readable by scripts (localStorage, useCookie or setCookie without httpOnly, document.cookie)

• N15: private runtimeConfig secrets read in components, pages or composables

• N16: v-html with API data, Markdown output or a value whose source the scanner cannot see

What's included

• The scanner (Python 3.9+, standard library only)

• A rules guide with the reason and safer approach for each rule

• A risky and a safe example Nuxt project (nuxt.config.ts, .env.example, pages, a component, a composable and server routes), and the real report the scanner produces on the risky one

Limits

It reads each file on its own: values passed as props or through stores, checks inside helpers in other files, settings from Nuxt layers or modules, and the built .output folder are not followed. The nuxt-security module, Content Security Policy headers and dependency vulnerabilities are not checked. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 31 files: SKILL.md, the scanner in scripts/ (6 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (22 example files plus report.md, the real scanner output on the risky example).

How to use: This is an agent skill: a folder with a SKILL.md file and Python scripts that an AI coding agent such as Claude Code, Cursor, or Codex CLI can run. Unzip it into your agent's skills folder (for Claude Code: ~/.claude/skills/), then ask your agent to check your project. The scripts also run on their own with Python 3.9 or newer, with nothing to install. They read your files only: no network access, and nothing is sent anywhere.