DOCUMENTATION

Quickstart

Mayil is a web app, not a CLI. Paste a URL into the free-audit form on the homepage — no signup, no card — and get a scored, prioritized report back in seconds. Under the hood, that form calls the same POST /api/audit endpoint you can call directly.

Request

POST /api/audit
{ "url": "https://example.com/some-page" }

Response (trimmed)

{
  "auditId": "b3f1…",
  "audit": {
    "url": "https://example.com/some-page",
    "status": "complete",
    "scoreVersion": "v2",
    "score": {
      "overall": 72,
      "grade": "C",
      "categories": [ { "category": "answerFormat", "score": 61 }, … ],
      "groups": [ { "id": "eeat", "kind": "construct", "score": 68, "weight": 5 }, … ],
      "capped": null
    },
    "findings": [ { "checkId": "h1-present", "status": "pass", "score": 1, "weight": 3, … } ],
    "recommendations": [ { "checkId": "direct-answer", "scoreImpact": 4, … } ]
  }
}

That shape is the real Audit object the engine returns — score, per-check findings, and ranked recommendations, all in one response. There is no separate install step and no command-line tool.

Scoring model

Every audit produces a single 0–100 overall score with a letter grade. Each atomic check contributes a pass / warn / fail finding weighted by severity, blended into either one of the seven display categories below or one of six cross-cutting constructs:

E-E-A-T — Experience · Expertise · Authoritativeness · TrustJTBD — Jobs To Be DoneYMYL — Your Money or Your LifeAI Access & Crawlability — Can an AI engine reach and parse the page at all?Machine Readability — Can it cleanly extract the content and data once reached?Citeworthiness — Is there something on the page worth quoting?

YMYL is conditional: it only takes over as a check's scoring parent when the page covers a health, financial, legal, or safety topic, and it escalates that check's weight when it does.

Status meaning

PASSCheck met — contributes its full weighted score.

WARNPartially met — a reduced score, flags something worth fixing.

FAILNot met — contributes zero for that check's weight.

Checks that don't apply to a given page — or that error out, or can't be determined — carry weight 0 and are excluded from the score denominator entirely. They never silently drag a page down.

A failed critical check — for example, blocking AI crawlers outright in robots.txt — caps the overall score at a fixed ceiling (e.g. 20), no matter how strong every other signal is. That gate only ever lowers a high score; it never raises a low one.

The 7 dimensions

Every check belongs to exactly one of seven display categories. 130+ atomic checks ship today, each a small, evidence-backed unit — a status, a score contribution, and a snippet of what was actually found on the page.

Technical SEO

Whether the page is reachable and indexable at all — status codes, HTTPS, canonicals, robots directives.

http-status2xx response, no dead end

https-hstsServed over HTTPS with HSTS

canonicalA single canonical URL declared

Content Structure

Whether the HTML is organized so a parser (not just a human eye) can find the hierarchy.

h1-presentExactly one descriptive H1

heading-orderNo skipped heading levels

internal-linksMeaningful internal link graph

AI Accessibility

Whether AI crawlers specifically (GPTBot, ClaudeBot, PerplexityBot) are allowed in and can extract the content cheaply.

ai-crawler-accessrobots.txt does not block AI bots

content-extractabilityMain content is extractable, not JS-locked

llms-txt-qualityllms.txt present and well-formed

Content Quality

Whether the writing itself is substantive, original, and worth an engine repeating.

content-depthSubstantive, non-thin coverage

citation-worthinessSpecific, quotable claims

original-dataFirst-party data or research

Answer Format

Whether an answer engine can lift a self-contained answer without rewriting the page.

direct-answerA direct answer near the top

faq-contentExtractable Q&A pairs

question-h2sSubheadings phrased as questions

Authority

Whether a real, identifiable, credentialed source stands behind the page.

author-bylineNamed author on the page

author-credentialsVerifiable author expertise

entity-consistencyConsistent brand/entity identity

Structured Data

Whether machine-readable schema.org markup makes the page and its entities explicit.

json-ld-presentValid JSON-LD on the page

article-schemaArticle/BlogPosting schema

org-schemaOrganization schema for the brand

Full methodology, the pipeline diagram, and all six cross-cutting constructs (E-E-A-T, JTBD, YMYL, AI Access, Machine Readability, Citeworthiness) in more depth: How it works →

Partner API

There is no public REST API with arbitrary endpoints. What exists is a stable, versioned JSON export — toPartnerApi(audit) — that reshapes an Audit into a flatter payload intended for partners piping results into their own dashboards:

{
  "apiVersion": "1",
  "scoreVersion": "v2",
  "url": "https://example.com/some-page",
  "auditedAt": "2026-07-07T00:00:00.000Z",
  "status": "complete",
  "score": { "overall": 72, "grade": "C", "capped": null, "groups": [ … ] },
  "findings": [ { "checkId": "h1-present", "status": "pass", "score": 1,
      "scoringParent": { "kind": "category", "id": "contentStructure" },
      "facets": [], "severity": "high", "evidence": { "count": 1 }, "lifecycle": "active" } ],
  "recommendations": [ { "checkId": "direct-answer", "title": "…", "scoreImpact": 4, … } ]
}

For the paid-tier, full-site brand-authority audit, the async job is polled at GET /api/site-audit/:jobId, which returns one of:

{ "status": "pending" }
{ "status": "complete", "audit": { /* SiteAudit */ } }
{ "status": "failed", "error": "…" }

That route is ownership-guarded: a job tied to a signed-in organization is only visible to that organization; anonymous jobs remain publicly readable by their link. There is no self-serve API-key system today — partner access is arranged directly.