ai-seo-audit is a free, open-source web app that audits a page for AI answer-engine readability — the kind of optimization that gets you cited by ChatGPT, Claude, and Perplexity rather than ranked on Google. Paste a URL, get a 0–100 score across six categories (crawlability, content chunking, schema, E-E-A-T, off-site citations, and governance), then copy a generated fix prompt straight into your coding agent.

Sometimes called GEO (Generative Engine Optimization) or AEO (Answer Engine Optimization).

Paste a URL → get a score in ~30 seconds → copy the fix prompt into Claude Code, Cursor, or any agent.

┌──────────────────────────────────────────────────┐
│  https://example.com/post                        │  →  Score: 72 / 100  (Strong)
└──────────────────────────────────────────────────┘
   ├─ A. Technical AI Crawlability ..... 18.0 / 28.4
   ├─ B. Content Structure & Chunking .. 21.0 / 25.4
   ├─ C. Structured Data / Schema ...... 13.4 / 13.4
   ├─ D. E-E-A-T & Entity Authority ... 13.0 / 21.4
   ├─ E. Off-site / Citation Surface ... 3.6 / 8.4
   └─ F. Measurement & Governance ...... 3.0 / 3.0

• 30+ AI-readability checks across crawlability, content chunking, schema markup, and trust signals.
• Agent-ready fix prompt — generated per audit, paste into Claude Code / Cursor / Aider to implement the fixes.
• JSON API at /api/audit?domain=... for CI pipelines and dashboards.
• Edge-cached results via optional Turso (libSQL) — repeat audits are instant.
• Zero direct page fetches. All web data flows through context.dev, giving you clean Markdown extraction and rendered HTML without running a headless browser yourself.

flowchart LR
  user([User])
  subgraph next["Next.js 16 App (Vercel)"]
    ui["UI<br/><sub>app/, components/</sub>"]
    api["API<br/><sub>app/api/audit/route.ts</sub>"]
    audit["Audit orchestrator<br/><sub>lib/audit.ts</sub>"]
    rules["Rule engine<br/><sub>lib/rules.ts (30+ rules)</sub>"]
    prompt["Agent fix prompt<br/><sub>lib/agent-fix-prompt.ts</sub>"]
  end
  ctx[("context.dev<br/><sub>Markdown + HTML scrape</sub>")]
  turso[("Turso / libSQL<br/><sub>optional cache</sub>")]
  target[("Target page")]

  user -->|paste URL| ui
  ui --> api
  api --> audit
  audit -->|webScrapeMd / webScrapeHTML| ctx
  ctx -->|fetch| target
  audit --> rules
  rules --> audit
  audit --> prompt
  audit <-->|"cache (sha256 key)"| turso
  audit -->|"score + items + prompt"| api
  api -->|JSON| ui

The app never hits the audited site directly — every byte of page content is fetched through context.dev, which handles rendering, robots compliance, and Markdown extraction.

Context.dev handles the page fetch + Markdown extraction that powers the audit. It has a generous free tier.

👉 Sign up for context.dev

After signing in, grab your API key from the dashboard.

git clone https://github.com/context-dot-dev/ai-seo-audit.git
cd ai-seo-audit
npm install

Copy the example file and fill in your key:

cp .env.example .env

# .env
CONTEXT_DEV_API_KEY=ctx_sk_...          # required — from https://context.dev
NEXT_PUBLIC_SITE_URL=http://localhost:3000

# Optional: Turso (libSQL) cache. Without these, audits run live every time.
aiseo_TURSO_DATABASE_URL=
aiseo_TURSO_AUTH_TOKEN=

Turso is optional. When the Turso variables are missing the app falls back to running a fresh audit on every request. For production, the Vercel + Turso Marketplace integration wires both vars in automatically.

npm run dev

Open http://localhost:3000 and paste a URL.

npm run build
npm run start

Visit /, paste a URL, hit Audit. Results live at /audit/<domain> and are shareable.

# GET
curl "http://localhost:3000/api/audit?domain=example.com"

# POST
curl -X POST http://localhost:3000/api/audit \
  -H "content-type: application/json" \
  -d '{"domain": "example.com"}'

Response shape (abbreviated):

{
  "status": "ready",
  "cached": false,
  "audit": {
    "url": "https://example.com/",
    "score": 72,
    "band": { "label": "Strong", "interpretation": "..." },
    "categories": [
      {
        "id": "A",
        "name": "Technical AI Crawlability",
        "score": 18.3,
        "maxScore": 28.4,
        "items": [
          /* per-rule */
        ],
      },
      {
        "id": "B",
        "name": "Content Structure & Chunking",
        "score": 21,
        "maxScore": 25.4,
      },
    ],
    "topPriorities": ["A11: Add a self-referencing canonical tag ...", "..."],
    "agentPrompts": { "full": "Paste this into your coding agent ..." },
  },
}

• Next.js 16 (App Router, React 19, server components)
• Tailwind CSS v4
• context.dev SDK for page fetching + Markdown extraction
• Turso (libSQL) for optional edge-cached results
• TypeScript end-to-end

ai-seo-audit/
├─ app/
│  ├─ api/audit/route.ts        # JSON API
│  ├─ audit/[domain]/page.tsx   # Shareable results page
│  ├─ page.tsx                  # Landing + audit form
│  └─ layout.tsx, robots.ts, sitemap.ts
├─ components/
│  ├─ audit-form.tsx
│  └─ audit-results-view.tsx
├─ lib/
│  ├─ audit.ts                  # Orchestrator (fetch → parse → score)
│  ├─ perform-audit.ts          # Caching + error envelope
│  ├─ rules.ts                  # Every audit rule lives here
│  ├─ agent-fix-prompt.ts       # Builds the copy-paste fix prompt
│  ├─ turso.ts                  # Cache (libSQL)
│  └─ url.ts, brand.ts, ...
└─ public/                      # Favicons, OG images

All audit logic lives in lib/rules.ts. To add a check:

1. Append an entry to the RULES array.
2. Point its categoryId at an existing entry in CATEGORIES.
3. Implement evaluate(ctx: RuleContext) returning { status: 'pass' | 'partial' | 'fail' | 'na', evidence }.
4. Optionally set multiplier to weight the rule inside its category.

Scoring, the UI breakdown, and the agent fix prompt all derive from this file — no other code needs to change.

The fastest path is Vercel + Turso Marketplace:

1. Click Import in Vercel, point it at your fork.
2. Add CONTEXT_DEV_API_KEY from context.dev under Project Settings → Environment Variables.
3. Add the Turso integration from the Vercel Marketplace — it provisions aiseo_TURSO_DATABASE_URL and aiseo_TURSO_AUTH_TOKEN for you.
4. Deploy.

Self-hosting works too — it's a standard Next.js app, runs anywhere Node 20+ runs.

Issues and PRs are welcome. Good first contributions:

• New audit rules in lib/rules.ts (cite a source for the heuristic in the PR description).
• Better evidence strings so the fix prompt gives agents more to work with.
• Bug reports with the URL audited + the score / category that surprised you.

Before opening a PR:

npm run lint
npm run typecheck
npm run build

MIT © Context.dev