set-up-security-headers
/set-up-security-headers
Use when hardening a frontend SPA's delivery — one strict Content-Security-Policy and header set (no unsafe-inline scripts), rendered for the project's host (Netlify, Vercel, or self-hosted Caddy/nginx/Traefik), plus Dependabot hygiene and an honest note on the XSS surface.
- Author
- velimirmueller ↗
- Repository
- ai-development-skills ↗
- Licence
- MIT
- Last repository change
- 23.10.2026
Assessment and security
- Files in this package
- 2
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 9
No signals detected. Repository changed on 23.10.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Use when hardening a frontend SPA's delivery — one strict Content-Security-Policy and header set (no unsafe-inline scripts), rendered for the project's host (Netlify, Vercel, or self-hosted Caddy/nginx/Traefik), plus Dependabot hygiene and an honest note on the XSS surface.
Original author description
Use when hardening a frontend SPA's delivery — one strict Content-Security-Policy and header set (no unsafe-inline scripts), rendered for the project's host (Netlify, Vercel, or self-hosted Caddy/nginx/Traefik), plus Dependabot hygiene and an honest note on the XSS surface.
How it works
- 1. Audit current state
- 2. Decide
- 3. The policy — one definition, every host
- 4. Render it for the host (a project uses exactly one block)
- 5. CSP notes for a Vite SPA
- 6. Dependency hygiene
- 7. The XSS surface (the honest note)
- 8. Verify
- References
SKILL.md preview
---
name: set-up-security-headers
description: Use when hardening a frontend SPA's delivery — one strict Content-Security-Policy and header set (no unsafe-inline scripts), rendered for the project's host (Netlify, Vercel, or self-hosted Caddy/nginx/Traefik), plus Dependabot hygiene and an honest note on the XSS surface.
---
# Set Up Security Headers
Security headers are delivered by the host, not the bundle. The policy is the same on every host; only the file that carries it changes. This skill writes the policy once (step 3), renders it for the project's host (step 4), and wires `connect-src` to the origins the app actually calls.
## 1. Audit current state
```bash
cat .claude/stack-profile.md 2>/dev/null || cat ~/.claude/stack-profile.md 2>/dev/null # hosting, ci
ls netlify.toml public/_headers vercel.json vercel.ts Caddyfile nginx.conf compose*.y*ml .github/dependabot.yml 2>/dev/null
grep -in "content-security-policy\|strict-transport" netlify.toml public/_headers vercel.json vercel.ts Caddyfile nginx.conf 2>/dev/null
grep -n "VITE_API_URL\|VITE_REALTIME_URL" src/libs/env.ts 2>/dev/null # origins for connect-src
grep -n "<script>" index.html 2>/dev/null # inline scripts break script-src 'self'
```
Pick the host branch from the profile's `hosting` list. If there is no profile, detect it from the files above.
| Profile `hosting` / evidence | Branch |
|---|---|
| `netlify`, or `netlify.toml` | Netlify (4a) |
| `vercel`, or `vercel.json` / `vercel.ts` | Vercel (4b) |
| `hetzner`, `ionos`, or a Caddyfile / nginx conf / Traefik labels | Self-hosted (4c) |
| `supabase`, `none`, or unknown | Ask one question: where is the SPA served from? Do not guess. |
**Prerequisites:** a deployed host (see `configure-ci`). Recommended: `validate-env` (the CSP reads `VITE_API_URL` / `VITE_REALTIME_URL`).
## 2. Decide
- No headers → full setup. Partial → add the missing headers/directives. Present → confirm `connect-src` covers current API + realtime origins.
- An inline `<script>` in `index.html` (theme init, analytics snippet) → move it to a same-origin file in `public/` first. The policy below blocks inline scripts on purpose.
## 3. The policy — one definition, every host
| Header | Value |
|---|---|
| `Content-Security-Policy` | `default-src 'self'; script-src 'self'; style-src 'self' 'unsafe-inline'; img-src 'self' data:; font-src 'self'; connect-src 'self' https://api.example.com wss://realtime.example.com; frame-ancestors 'none'; base-uri 'self'; form-action 'self'; object-src 'none'` |
| `Strict-Transport-Security` | `max-age=31536000; includeSubDomains` (add `; preload` only after the HSTS-preload decision in `security-patterns.md`) |
| `X-Content-Type-Options` | `nosniff` |
| `Referrer-Policy` | `strict-origin-when-cross-origin` |
| `Permissions-Policy` | `camera=(), microphone=(), geolocation=()` |
Replace `https://api.example.com` / `wss://realtime.example.com` with the **origins** of `VITE_API_URL` / `VITE_REALTIME_URL` (scheme + host, no path). Drop the `wss:` origin if realtime isn't set up. Cache rule for every host: `index.html` and the service worker (`sw.js`) must not be cached long-term (`Cache-Control: no-cache`); hashed `/assets/*` can be `immutable`. The SPA fallback (all unknown paths → `/index.html`) is host config too; each block below includes it.
## 4. Render it for the host (a project uses exactly one block)
### 4a. Netlify
`netlify.toml` — the `[[headers]]` block (`configure-ci` owns `[build]`; merge, don't overwrite):
```toml
[[headers]]
for = "/*"
[headers.values]
Content-Security-Policy = "default-src 'self'; script-src 'self'; style-src 'self' 'unsafe-inline'; img-src 'self' data:; font-src 'self'; connect-src 'self' https://api.example.com wss://realtime.example.com; frame-ancestors 'none'; base-uri 'self'; form-action 'self'; object-src 'none'"
Strict-Transport-Security = "max-age=31536000; includeSubDomains"
X-Content-Type-Options = "nosniff"
Referrer-Policy = "strict-origin-when-cross-origin"
Permissions-Policy = "camera=(), microphone=(), geolocation=()"
[[headers]]
for = "/sw.js"
[headers.values]
Cache-Control = "no-cache"
```
### 4b. Vercel
`vercel.json` (use `vercel.ts` with `routes.header()` from `@vercel/config/v1` only if the project already has one — a project may hold one config file, not both):
```json
{
"$schema": "https://openapi.vercel.sh/vercel.json",
"rewrites": [{ "source": "/(.*)", "destination": "/index.html" }],
"headers": [
{
"source": "/(.*)",
"headers": [
{ "key": "Content-Security-Policy", "value": "default-src 'self'; script-src 'self'; style-src 'self' 'unsafe-inline'; img-src 'self' data:; font-src 'self'; connect-src 'self' https://api.example.com wss://realtime.example.com; frame-ancestors 'none'; base-uri 'self'; form-action 'self'; object-src 'none'" },
{ "key": "Strict-Transport-Security", "value": "max-age=31536000; includeSubDomains" },
{ "key": "X-Content-Type-Options", "value": "nosniff" },
{ "key": "Referrer-Policy", "value": "strict-origin-when-cross-origin" },
{ "key": "Permissions-Policy", "value": "camera=(), microphone=(), geolocation=()" }
]
},
{ "source": "/sw.js", "headers": [{ "key": "Cache-Control", "value": "public, max-age=0, must-revalidate" }] }
]
}
```
Preview deploys and project setup: [`../../infra/deploy-to-vercel/SKILL.md`](../../infra/deploy-to-vercel/SKILL.md). Vercel serves files from the filesystem before it applies the rewrite, so `/assets/*` still resolves.
### 4c. Self-hosted (Hetzner / IONOS)
Caddy serves the built `dist/` and carries the headers (a Caddy container behind or instead of Traefik):
```caddyfile
example.com {
root * /srv
encode zstd gzip
header {
Content-Security-Policy "default-src 'self'; script-src 'self'; style-src 'self' 'unsafe-inline'; img-src 'self' data:; font-src 'self'; connect-src 'self' https://api.example.com wss://realtime.example.com; frame-ancestors 'none'; base-uri 'self'; form-action 'self'; object-src 'none'"
Strict-Transport-Security "max-age=31536000; includeSubDomains"
X-Content-Type-Options nosniff
Referrer-Policy strict-origin-when-cross-origin
Permissions-Policy "camera=(), microphone=(), geolocation=()"
}
@shell path /index.html /sw.js
header @shell Cache-Control "no-cache"
try_files {path} /index.html
file_server
}
```
nginx: the same values as `add_header <name> "<value>" always;` in the `server` block. An `add_header` inside a `location` block **replaces** every inherited one — repeat the full set there or use an `include`. Traefik in front of a plain file server: use the `headers` middleware (`stsSeconds=31536000`, `stsIncludeSubdomains=true`, `contentTypeNosniff=true`, `referrerPolicy`, `contentSecurityPolicy`, `customResponseHeaders.Permissions-Policy`) and attach it to the router. Server, deploy and TLS: [`../../infra/deploy-to-hetzner/SKILL.md`](../../infra/deploy-to-hetzner/SKILL.md).
## 5. CSP notes for a Vite SPA
- Vite emits **external** scripts in prod, so `script-src 'self'` needs no `'unsafe-inline'`. The dev server is looser; the policy applies to the deployed build.
- `style-src 'unsafe-inline'` is a pragmatic default, not a requirement: the Tailwind build is an external CSS file, and React/Vue set `style` through the CSSOM, which CSP allows. Libraries that inject `<style>` tags at runtime need it. Test `style-src 'self'` in Report-Only first (`security-patterns.md`).
- `connect-src` breaks silently: omit your API or `wss://` origin and those requests fail with a console CSP error. That's why it reads from the env seam.
- Trusted Types (`require-trusted-types-for 'script'`) is the next layer. Roll it out in Report-Only (`security-patterns.md`).
## 6. Dependency hygiene
`.github/dependabot.yml`:
```yaml
version: 2
updates:
- package-ecosystem: npm
directory: "/"
schedule:
interval: weekly
groups:
minor-and-patch:
update-types: [minor, patch]
- package-ecosystem: github-actions
directory: "/"
schedule:
interval: weekly
```
`configure-ci` runs `pnpm audit`. Automation proposes; a human approves each bump.
## 7. The XSS surface (the honest note)
React and Vue escape interpolated values by default. The real hole is **`dangerouslySetInnerHTML` (React) / `v-html` (Vue)** — they bypass escaping. Avoid them; if unavoidable, sanitize with DOMPurify first. A strict CSP is defense-in-depth, not a substitute. Baseline: [`security-baseline.md`](../../core/_shared/security-baseline.md).
## 8. Verify
```bash
curl -sI https://<deploy-url>/ | grep -i "content-security-policy\|strict-transport\|x-content-type"
curl -sI https://<deploy-url>/sw.js | grep -i cache-control # if the PWA is on
curl -s -o /dev/null -w "%{http_code}\n" https://<deploy-url>/some/deep/route # 200 = SPA fallback works
```
Load the app: the console shows **no** CSP violations (every API / realtime / font / image origin is allowed). Grade it at securityheaders.com.
## References
- ./security-patterns.md — edge-vs-meta delivery, the `connect-src`-from-env rule, strict-CSP rollout (Report-Only, Trusted Types), the HSTS-preload decision, supply chain, the XSS surface.
- ../validate-env/SKILL.md — the `env` seam supplying the `connect-src` origins.
- ../configure-ci/SKILL.md — owns the Netlify `[build]` block and the `pnpm audit` step.
- ../../core/_shared/security-baseline.md — the cross-stack header and cookie rules this skill implements.
Files in this package (2)
- SKILL.md 9 KB
- security-patterns.md 6 KB
Versions
The database has no other skills named set-up-security-headers.
Database snapshot: 13.11.2026 · package version f33d7c776884 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100security
Security rules for projects following the air-gapped, declarative-first architecture. Covers Content Security Policy, recommended security headers, and the architectural decisions that enable a strict CSP without unsafe-inline or unsafe-eval. Use when configuring deployment,… - 95memstack-security-csp-headers
Use this skill when the user says 'CSP', 'Content-Security-Policy', 'security headers', 'HSTS', 'X-Frame-Options', 'clickjacking', 'unsafe-inline', 'unsafe-eval', or needs to audit, generate, or fix HTTP security headers for a web application. Do NOT use for API route audits or dependency scanning. - 100content-security-policy-headers
Use when designing or fixing a Content Security Policy on a real site, choosing between nonce-based and hash-based CSP, adding strict-dynamic, debugging "Refused to execute inline script" errors, deploying CSP in report-only mode first, configuring report-to / report-uri, or… - 100vercel-security-basics
Apply Vercel security best practices for secrets, headers, and access control. Use when securing API keys, configuring security headers, or auditing Vercel security configuration. Trigger with phrases like "vercel security", "vercel secrets", "secure vercel", "vercel headers", "vercel CSP". - 100security-headers-configuration
Configures HTTP security headers to protect against XSS, clickjacking, and MIME sniffing attacks. Use when hardening web applications, passing security audits, or implementing Content Security Policy. - 90frontend-security
Use this skill when the user says 'XSS', 'CSP', 'CSRF', 'SRI', 'secure cookies', 'security headers', 'frontend security', 'sanitize input', 'content security policy', 'cross-site scripting', 'cross-site request forgery', or when auditing frontend security. This skill enforces:…