AI Citability Check (ai-citability-check)

Summary

Passage-level scoring for AI citation readiness across sampled pages.

Classification

• Category: SEO
• Plugin id: ai-citability-check
• Version: 1.0.0
• Target types: WEBSITE, WEBSHOP
• Weight class: LIGHT
• Risk level: not set on plugin object
• Browser required: No (HTTP / Node / other — see source)
• Tier / group: Tier 1 — HTTP-only

Schedule

• scheduleHints: { defaultInterval: '0 5 * * 1', priority: 4 }
• Prerequisites: None declared on this plugin.

Configurable inputs

Public keys are derived from inputSchema: z.object({ ... }) (underscore-prefixed keys are runtime-only and omitted here).

Field	Notes
timeout	Zod field in inputSchema — see source for defaults, min/max, and .describe().
maxPages	Zod field in inputSchema — see source for defaults, min/max, and .describe().

Runtime / injected config

• None documented beyond standard worker context (target, config, logger, reportProgress). Underscore-prefixed keys are internal.

Checks that produce findings

These checks run inside this plugin during a single test run. They are not separate tests in the database — each becomes a Finding linked to the Run for ai-citability-check.

Check	Typical severity	Why
No substantial content blocks found	WARNING	No blocks with ≥20 words after stripping nav/footer — nothing for AI systems to quote.
Very low overall citability score (< 40/100 avg)	WARNING	Heuristic score across dimensions suggests content is unlikely to be cited in AI answers.
Moderate citability (40–59 avg)	INFO	Some extractable passages exist but many blocks lack AI-preferred structure.
Content lacks answer-first patterns	WARNING	Average Answer Block Quality < 10/30 — sections rarely open with definitions or direct facts.
Content lacks statistics and data points	INFO	Average Statistical Density < 4/15 — few percentages, amounts, or named sources for models to cite.
Content blocks are not self-contained	INFO	Average Self-Containment < 8/25 — heavy pronoun use or wrong length for standalone quotes.
No passages in optimal AI citation length (134–167 words)	INFO	Zero blocks in the research-backed citation length band when ≥3 blocks exist.

How each check works

1. Sample pages — Fetch homepage; if maxPages > 1 (default 3), discover same-host links prioritizing paths matching blog/article/guide/about/services/product, up to maxPages − 1 extra URLs.
2. Extract blocks — Remove script/style/nav/footer/header/aside/forms; split on headings; merge paragraphs until block has ≥20 words.
3. Score each block (0–100) — Five weighted dimensions: Answer Block Quality (30), Self-Containment (25), Structural Readability (20), Statistical Density (15), Uniqueness Signals (10). Supports English and Norwegian patterns (e.g. “X er en”, “ifølge”, “vi fant”).
4. Grades — A≥80, B≥65, C≥50, D≥35, F below; stored in metrics, not separate findings.
5. Aggregate findings — Thresholds above on averages and optimal-length count (optimalPassages === 0 and allScores.length >= 3).
6. Run score — Equals average citability score across all blocks.

Caveats and concerns

• Heuristic, not LLM-judged — Scores are regex/structure rules, not predictions of actual ChatGPT/Perplexity citations.
• English/Norwegian bias — Definition and statistic patterns favor these languages.
• DOM cleanup — Sites with content outside main/article/p may yield few blocks.
• No per-block findings — Only site-level aggregates; weak passages are listed in evidence on some findings.
• Short pages — Marketing one-liners may score poorly by design.

Source

packages/test-plugins/src/plugins/ai-citability-check.ts