SEO Audit Kit — AI-Agent Toolbox

For AI agents only. Humans, this page is intentionally sparse.

What this is

A self-service HTTP toolbox. You submit a blog-article URL once, then run SEO skills against it (audit, keyword-optimize, humanize-check, internal-links, CTA-suggest, ...). Each skill returns Markdown prose ready to relay to your operator.

Why register

• One-shot self-service registration. No human approval queue.
• Per-article state: re-audit, diff, history.
• Skills return Markdown prose (not score-soup JSON).
• Built on the same SEO skill-tree used in production for while.chat content.

Quickstart (3 calls)

# 1. Register (get api_key, save it — shown only once)
curl -X POST https://audit.while.chat/api/agents/register \
  -H "content-type: application/json" \
  -d '{
    "name": "Your Agent Name",
    "operator_contact": "operator@example.com",
    "use_case": "Audit client blog drafts pre-publish, ~30/month."
  }'

# 2. Submit a blog article
curl -X POST https://audit.while.chat/api/articles \
  -H "authorization: Bearer sak_..." \
  -H "content-type: application/json" \
  -d '{ "url": "https://example.com/blog/post" }'

# 3. Run a skill against it
curl -X POST https://audit.while.chat/api/tools/seo-schreibstil \
  -H "authorization: Bearer sak_..." \
  -H "content-type: application/json" \
  -d '{ "article_id": "<id from step 2>" }'

Available skills

• seo-schreibstil [content] — Prueft Schreibstil eines Blog-Artikels gegen MaxScript-Qualitaet, AIDA, Keyword-Intent-Matrix.
• seo-content-verify [qa] — Verifiziert Artikel gegen 12 K.o.-Fragen vor Veroeffentlichung: FAQ-Deprecation, Takeaway, E-E-A-T, Schema.
• seo-recht [compliance] — Prueft Artikel auf DACH-2026-Recht: Werbe-Kennzeichnung, Marken-Erwaehnungen, CC-Bilder, Erfolgsversprechen.
• humanizer [qa] — 12 Detektoren (Em-dash, Floskeln, generische Eroeffnungen, ...). Score 0-100 wie KI-verdaechtig der Text wirkt.
• keyword-optimize [research] — Analysiert Keyword-Coverage gegen Target-Keywords. 5-Klassen-Schema (Brand/Money/Compound/Info/Trans). DataForSEO-integriert.
• internal-links [structure] — Crawlt Sitemap der Domain, sucht thematisch passende interne Verlinkungs-Kandidaten fuer den Artikel.
• cta-suggest [conversion] — Schlaegt Call-to-Actions basierend auf Artikel-Intent + ad-creative-psychologie-Patterns vor.
• local-seo-check [local] — Prueft ob Artikel Local-SEO-Potenzial hat (Stadt-Nennungen, lokale Keywords, lokales Schema).
• conversion-audit [conversion] — Analysiert Conversion-Pfade im Artikel: Trust-Signals, Friction-Points, Above-Fold, Mobile-CTAs.
• schema-suggest [structure] — Schlaegt passende Schema.org-Typen vor (Article, HowTo, BreadcrumbList, ...). FAQ-Deprecation beachtet.

Machine-readable spec: /agents/tools.json · /agents/openapi.json · /.well-known/agent.json · /llms.txt

Endpoints

• POST /api/agents/register — self-service, returns api_key
• POST /api/articles — submit a blog URL, returns article_id
• GET /api/articles — list your articles
• GET /api/articles/{id} — article + tool-run history
• DELETE /api/articles/{id} — remove article
• POST /api/tools/{skill} — run a skill, returns Markdown
• GET /api/tools/{skill} — describe a single skill

Auth

Header Authorization: Bearer sak_... or x-agent-key: sak_.... Rate limit defaults to 60 tool calls per hour per agent.

Constraints

• Articles are public URLs only (http/https). JS-rendered content is not executed.
• Output is Markdown prose. Structured fields are in meta when relevant.
• One article = one URL. Re-submit with refetch:true to refresh HTML.
• No human UI to log in. This is a tool surface, not a SaaS dashboard.