@growthub/cli 0.14.4 → 0.14.6

package/assets/worker-kits/growthub-geo-seo-v1/docs/scoring-methodology.md

@@ -1,230 +0,0 @@
-# GEO Scoring Methodology
-
-**Source of truth for all scoring rules. The operator must apply these formulas exactly.**
-
----
-
-## GEO Score Formula
-
-The GEO Score is a weighted composite of 6 component scores, each ranging from 0 to 100.
-
-```
-GEO Score = (AI Citability & Visibility × 0.25)
-          + (Brand Authority × 0.20)
-          + (Content Quality & E-E-A-T × 0.20)
-          + (Technical Foundations × 0.15)
-          + (Structured Data × 0.10)
-          + (Platform Optimization × 0.10)
-```
-
-### Component Weights
-
-| Component | Weight | Rationale |
-|---|---|---|
-| AI Citability & Visibility | 25% | Core measure of AI search readiness — crawler access + citability signals |
-| Brand Authority | 20% | AI systems prefer to cite recognized, cross-platform brands |
-| Content Quality & E-E-A-T | 20% | Google and AI systems both weight E-E-A-T heavily for content ranking |
-| Technical Foundations | 15% | Technical barriers prevent all other optimization from working |
-| Structured Data | 10% | Schema markup directly feeds AI answer surfaces (Google AI Overviews, ChatGPT) |
-| Platform Optimization | 10% | Platform-specific readiness for the 4 major AI search engines |
-
-### Computation Rules
-
-- Each component score is 0–100 (no decimals before aggregation)
-- Apply weights before rounding
-- Round the composite score to the nearest integer
-- If a component score is unavailable (data gap), use **50** as the neutral default
-- Flag any data-gap component in the GeoScoreSummary output
-
-**Example:**
-```
-AI Citability:  58 × 0.25 = 14.50
-Brand Auth:     66 × 0.20 = 13.20
-Content:        71 × 0.20 = 14.20
-Technical:      84 × 0.15 = 12.60
-Schema:         42 × 0.10 =  4.20
-Platform:       61 × 0.10 =  6.10
-                            ─────
-Composite:                  64.80 → rounds to 65
-```
-
----
-
-## Letter Grade Thresholds
-
-| Grade | Score Range | AI Search Status |
-|---|---|---|
-| A | 85–100 | Highly optimized — strong citability, clean crawler access, rich schema |
-| B | 70–84 | Good — some gaps, addressable in one sprint cycle |
-| C | 55–69 | Moderate — missing key citability signals, schema gaps, crawler issues |
-| D | 40–54 | Poor — likely not capturing meaningful AI-referred traffic |
-| F | Below 40 | Not AI-search-ready — critical issues across multiple components |
-
----
-
-## Citability Algorithm
-
-The Citability Score (one of the inputs to AI Citability & Visibility) uses a 5-metric algorithm.
-
-### 5-Metric Breakdown
-
-| Metric | Weight | What It Measures | Scoring Method |
-|---|---|---|---|
-| Answer Block Quality | 30% | Do paragraphs contain complete, self-sufficient answers AI can quote verbatim? | Score each paragraph 0–10; average across all paragraphs |
-| Self-Containment | 25% | Can each paragraph be understood without reading surrounding context? | Pronoun-to-noun ratio; lower = better self-containment |
-| Structural Readability | 20% | Does the page use headings, short paragraphs, and lists to enable AI parsing? | Check H1/H2/list/paragraph-length signals |
-| Statistical Density | 15% | Does the page contain specific numbers, percentages, and data references? | Count data points per 1,000 words; optimal: 8–15 |
-| Uniqueness Signals | 10% | Does the content contain proprietary claims or data not found elsewhere? | Check for first-party research, original data, unique terminology |
-
-**Citability Score formula:**
-```
-Citability = (Answer Block Quality × 0.30)
-           + (Self-Containment × 0.25)
-           + (Structural Readability × 0.20)
-           + (Statistical Density × 0.15)
-           + (Uniqueness Signals × 0.10)
-```
-
----
-
-### Metric Scoring Rules
-
-#### Answer Block Quality (0–100)
-
-Evaluate each paragraph on 3 criteria:
-1. Subject is clearly stated (no pronoun-as-subject opener)
-2. Supporting evidence or data is present
-3. No unresolved pronoun references
-
-Score each paragraph:
-- All 3 criteria met: 10/10
-- 2 criteria met: 7/10
-- 1 criterion met: 4/10
-- None met: 1/10
-
-Average across all paragraphs. Multiply by 10 to get 0–100 score.
-
-**High-score example:**
-> "GPTBot (used by ChatGPT's Browse mode) is blocked in the site's robots.txt. This means ChatGPT cannot fetch and cite this page's content, even when users directly ask about topics the page covers."
-
-**Low-score example:**
-> "It blocks them from accessing the site. This causes problems because they can't see what's there."
-*(No subject named, no evidence, multiple unresolved pronouns)*
-
----
-
-#### Self-Containment (0–100)
-
-**Pronoun inventory:** it, they, this, that, these, those, he, she, we (when antecedent is not in same sentence)
-
-1. Count pronouns used as subjects or objects without a prior noun in the same sentence
-2. Count total noun references (named entities + common nouns)
-3. Pronoun-to-noun ratio = pronoun count / noun count
-
-**Scoring:**
-- Ratio < 0.15: 100
-- Ratio 0.15–0.25: 85
-- Ratio 0.25–0.35: 70
-- Ratio 0.35–0.50: 50
-- Ratio 0.50–0.65: 30
-- Ratio > 0.65: 10
-
-Also penalize for word count outside the 300–3,000 word range:
-- < 300 words: score cap at 40 (too thin to be self-contained on a topic)
-- 300–800 words: light penalty (-10)
-- 800–2,500 words: optimal range, no penalty
-- > 2,500 words: light penalty (-5) — tends toward context-dependent sprawl
-
----
-
-#### Structural Readability (0–100)
-
-Start at 100 and deduct:
-
-| Issue | Deduction |
-|---|---|
-| No H1 present | -30 |
-| Fewer than 2 H2 sections | -15 |
-| No numbered or bulleted lists | -15 |
-| Average paragraph length > 150 words | -15 |
-| Any paragraph > 300 words (wall of text) | -10 per occurrence, capped at -20 |
-| No visual separation between sections | -10 |
-
-Minimum score: 0.
-
----
-
-#### Statistical Density (0–100)
-
-Count data points per 1,000 words. Data points include:
-- Percentage figures (e.g., "47%," "3x increase")
-- Specific numbers with units (e.g., "65ms," "$2,800," "14 crawlers")
-- Named statistics (e.g., "200M weekly active users")
-- Year references as evidence (e.g., "Q1 2026 data shows...")
-
-**Scoring:**
-- < 2 data points per 1,000 words: 10
-- 2–5: 35
-- 5–8: 60
-- 8–15: 100 (optimal range)
-- 15–25: 85 (slightly over-cited — readability may suffer)
-- > 25: 60 (data overload — hard for AI to extract clean answers)
-
----
-
-#### Uniqueness Signals (0–100)
-
-| Signal | Points |
-|---|---|
-| First-party study or original research | 35 |
-| Proprietary data or named internal data source | 25 |
-| Original methodology with named process | 20 |
-| Unique branded terminology | 15 |
-| Non-generic competitive differentiation statement | 10 |
-| Named case study with specific results | 10 |
-
-Cap at 100. If total > 100, use 100.
-
----
-
-## Component Score Normalization
-
-Each subagent returns a raw 0–100 score. No normalization is needed — all scores are already on the same scale.
-
-Do not normalize scores before applying weights. Apply weights directly to the 0–100 values.
-
----
-
-## Score Interpretation for Client Communication
-
-### Grade A (85–100)
-> "Your site is well-positioned for AI-driven search. You are in the top tier for citability, crawler access, and content quality. We recommend ongoing monthly monitoring and targeted improvements to maintain this position as AI search evolves."
-
-### Grade B (70–84)
-> "Good foundation. You have established AI search presence, but specific gaps are limiting your ceiling. Targeted improvements in [lowest-scoring component] can push you into the A tier within 30–60 days."
-
-### Grade C (55–69)
-> "Moderate visibility. You are not capturing significant AI-referred traffic yet. A full remediation roadmap is recommended. Addressing the top 3 gaps typically produces a measurable score improvement within 30 days."
-
-### Grade D (40–54)
-> "Your site has significant AI search blindspots. AI systems may not be citing or recommending your content at all, even for queries you should rank for. Immediate technical and content remediation is required."
-
-### Grade F (below 40)
-> "Critical issues detected. AI crawlers may be blocked outright, or your content lacks the structural signals needed for AI citation. A full remediation engagement is required before any AI-referred traffic is possible."
-
----
-
-## Benchmark Context
-
-These benchmarks are based on geo-seo-claude audit data:
-
-| Benchmark | Score |
-|---|---|
-| Portfolio average (all audited sites) | 58 / 100 |
-| Top quartile (25th percentile from top) | 76 / 100 |
-| Top 10% | 85 / 100 |
-| Minimum for meaningful AI-referred traffic | ~65 / 100 |
-| Sites that actively appear in Perplexity citations | ~72+ / 100 |
-| Sites that appear in Google AI Overviews regularly | ~78+ / 100 |
-
-**llms.txt adoption rate:** Approximately 22% of audited sites have a valid `llms.txt` as of Q1 2026.

package/assets/worker-kits/growthub-geo-seo-v1/docs/subagent-dispatch.md

@@ -1,273 +0,0 @@
-# Subagent Dispatch — 5-Parallel Orchestration
-
----
-
-## Overview
-
-Phase 2 of the GEO audit runs 5 specialized subagents in parallel. Each subagent receives the Phase 1 fetch data as input and returns a component score (0–100) plus a findings list. The operator synthesizes all 5 results into the composite GEO Score.
-
----
-
-## When Phase 2 Triggers
-
-Phase 2 begins after Phase 1 fetch completes successfully.
-
-**Phase 1 required outputs (all must be present before dispatching):**
-
-- robots.txt contents (or confirmed 404)
-- llms.txt status (present/missing/malformed)
-- Rendered page HTML
-- HTTP response headers
-- Page word count and heading structure
-- JSON-LD structured data (or confirmed absent)
-
-If Phase 1 fails to fetch the target URL entirely (connection refused, server error), stop and report the error. Do not dispatch Phase 2 with no data.
-
-If Phase 1 produces partial data (e.g., llms.txt returns 404), dispatch Phase 2 anyway and note the missing data in the relevant subagent's output.
-
----
-
-## Subagent Definitions
-
-### 1. geo-ai-visibility
-
-**File:** `agents/geo-ai-visibility.md` (in fork)
-**Scope:** AI crawler access and citation format quality
-**GEO Score weight:** 25%
-
-**Inputs from Phase 1:**
-- robots.txt full text
-- llms.txt full text (or `null` if missing)
-- llms-full.txt status
-- X-Robots-Tag header value
-- Page canonical URL
-
-**What it evaluates:**
-
-| Signal | Method | Score Contribution |
-|---|---|---|
-| GPTBot access | Parse `User-agent: GPTBot` in robots.txt | High |
-| ClaudeBot access | Parse `User-agent: ClaudeBot` | High |
-| PerplexityBot access | Parse `User-agent: PerplexityBot` | High |
-| Google-Extended access | Parse `User-agent: Google-Extended` | High |
-| Bingbot access | Parse `User-agent: Bingbot` | Medium |
-| Applebot-Extended access | Parse `User-agent: Applebot-Extended` | Medium |
-| Anthropic-AI access | Parse `User-agent: anthropic-ai` | Medium |
-| cohere-ai access | Parse `User-agent: cohere-ai` | Medium |
-| Meta-ExternalFetcher access | Parse `User-agent: meta-externalagent` | Low |
-| YouBot access | Parse `User-agent: YouBot` | Low |
-| DuckAssistBot access | Parse `User-agent: DuckAssistBot` | Low |
-| Scrapy access | Parse `User-agent: Scrapy` | Low |
-| CCBot access | Parse `User-agent: CCBot` | Low |
-| ia_archiver access | Parse `User-agent: ia_archiver` | Low |
-| llms.txt present | File exists and returns 200 | High |
-| llms.txt valid format | No parsing errors | Medium |
-| llms-full.txt present | File exists | Medium |
-| Citation format quality | Clean canonical, no JS-only content walls | High |
-| X-Robots-Tag noai | Penalize heavily if present | Critical |
-
-**Output format:**
-```json
-{
-  "subagent": "geo-ai-visibility",
-  "score": 0-100,
-  "crawler_results": { "GPTBot": "allowed|blocked|partial", ... },
-  "llms_txt_present": true|false,
-  "llms_full_txt_present": true|false,
-  "top_findings": ["finding 1", "finding 2", "finding 3"]
-}
-```
-
----
-
-### 2. geo-content
-
-**File:** `agents/geo-content.md` (in fork)
-**Scope:** E-E-A-T signals, answer block quality, content self-containment
-**GEO Score weight:** 20%
-
-**Inputs from Phase 1:**
-- Rendered page HTML (full text)
-- Page word count
-- Heading hierarchy (H1, H2, H3 counts)
-- Any author markup found
-
-**What it evaluates:**
-- Experience signals (case studies, original data, first-person examples)
-- Expertise signals (author bylines, credentials, domain terminology depth)
-- Authoritativeness signals (external citations, authority domain references)
-- Trustworthiness signals (HTTPS confirmed in Phase 1, contact info, privacy policy presence)
-- Answer block quality (5-metric citability algorithm — see `docs/scoring-methodology.md`)
-
-**Output format:**
-```json
-{
-  "subagent": "geo-content",
-  "score": 0-100,
-  "eeat_scores": { "experience": 0-10, "expertise": 0-10, "authoritativeness": 0-10, "trustworthiness": 0-10 },
-  "citability_score": 0-100,
-  "top_findings": ["finding 1", "finding 2", "finding 3"]
-}
-```
-
----
-
-### 3. geo-platform-analysis
-
-**File:** `agents/geo-platform-analysis.md` (in fork)
-**Scope:** Readiness for 4 major AI search platforms
-**GEO Score weight:** 10%
-
-**Inputs from Phase 1:**
-- Page HTML structure
-- Schema markup found
-- llms.txt status (from geo-ai-visibility)
-- Content signals (from geo-content)
-
-**What it evaluates:**
-
-| Platform | Key Signals Checked |
-|---|---|
-| ChatGPT (Browse mode) | Clean URL, no crawler block, answer-block format, Bing indexability |
-| Perplexity AI | Direct URL crawlability, self-contained paragraphs, citation format |
-| Google AI Overviews | FAQPage schema, E-E-A-T signals, featured snippet optimization, structured data |
-| Gemini | Google entity graph signals, Knowledge Panel readiness, broad schema coverage |
-
-**Output format:**
-```json
-{
-  "subagent": "geo-platform-analysis",
-  "score": 0-100,
-  "platform_scores": {
-    "chatgpt": 0-100,
-    "perplexity": 0-100,
-    "google_ai_overviews": 0-100,
-    "gemini": 0-100
-  },
-  "top_findings": ["finding 1", "finding 2", "finding 3"]
-}
-```
-
----
-
-### 4. geo-schema
-
-**File:** `agents/geo-schema.md` (in fork)
-**Scope:** Structured data coverage and validation
-**GEO Score weight:** 10%
-
-**Inputs from Phase 1:**
-- All JSON-LD blocks from page source
-- Microdata and RDFa markup (if present)
-- Page type determination
-
-**What it evaluates:**
-- Which schema types are present
-- Required property completeness for each type
-- Validation errors (missing required fields, incorrect value types)
-- Missing recommended schema types for the detected page type
-- Schema richness score (types × completeness)
-
-**Output format:**
-```json
-{
-  "subagent": "geo-schema",
-  "score": 0-100,
-  "types_found": ["Organization", "WebSite"],
-  "types_missing": ["FAQPage", "Article", "BreadcrumbList"],
-  "validation_errors": [
-    { "type": "Organization", "property": "contactPoint", "error": "missing required property" }
-  ],
-  "top_findings": ["finding 1", "finding 2", "finding 3"]
-}
-```
-
----
-
-### 5. geo-technical
-
-**File:** `agents/geo-technical.md` (in fork)
-**Scope:** Technical SEO and server health
-**GEO Score weight:** 15%
-
-**Inputs from Phase 1:**
-- Full HTTP response headers
-- HTTPS status
-- robots.txt validity
-- sitemap.xml accessibility
-- Page HTML (for Core Web Vitals structural signals)
-
-**What it evaluates:**
-
-| Signal | Check Method | Weight |
-|---|---|---|
-| HTTPS enforced | HTTP→HTTPS redirect + HSTS header | High |
-| HSTS present | `Strict-Transport-Security` header | Medium |
-| Mobile viewport | `<meta name="viewport">` present | Medium |
-| robots.txt valid syntax | Parser runs without errors | High |
-| Sitemap in robots.txt | `Sitemap:` directive present | Medium |
-| sitemap.xml accessible | HTTP 200 response | Medium |
-| llms.txt accessible | HTTP 200 response | High |
-| Canonical tag present | `<link rel="canonical">` in source | Medium |
-| Render-blocking scripts | `<script>` in `<head>` without defer/async | Medium |
-| Images have dimensions | `width` and `height` attributes | Low |
-| Content-Encoding | gzip or brotli present | Low |
-| Cache-Control | public caching enabled | Low |
-
-**Output format:**
-```json
-{
-  "subagent": "geo-technical",
-  "score": 0-100,
-  "https_enforced": true|false,
-  "hsts_present": true|false,
-  "mobile_ready": true|false,
-  "robots_valid": true|false,
-  "sitemap_accessible": true|false,
-  "critical_issues": ["issue 1", "issue 2"],
-  "top_findings": ["finding 1", "finding 2", "finding 3"]
-}
-```
-
----
-
-## Collecting Results
-
-After all 5 subagents return, the operator:
-
-1. Validates that all 5 responses are present
-2. Extracts the `score` from each response
-3. Applies the GEO Score formula (see `docs/scoring-methodology.md`)
-4. Collects all `top_findings` arrays into a unified findings list
-5. Ranks findings by impact for the Remediation Roadmap
-
----
-
-## Error Handling
-
-If a subagent fails or returns incomplete data:
-
-| Scenario | Action |
-|---|---|
-| Subagent returns no score | Use 50 as neutral default. Flag component as `data-gap` in GeoScoreSummary. |
-| robots.txt fetch fails (404/500) | Mark all crawler statuses as "unknown (robots.txt inaccessible)". Do not assume blocked or allowed. |
-| llms.txt returns 500 error | Treat as "missing" rather than "blocked". Note the error in CrawlerAccessReport. |
-| Page fetch fails entirely | Stop Phase 2. Report to user. Do not produce scores. |
-| One subagent times out | Log the timeout. Use 50 as default for that component. Note in output. |
-
-Never invent scores. A `data-gap` flag is more useful than a fabricated number.
-
----
-
-## Brand Authority (Separate Pass)
-
-Brand Authority (20% weight) is not a Phase 2 parallel subagent — it is a separate `/geo brands` command pass that runs `scripts/brand_scanner.py` against the 8 target platforms.
-
-Run it after Phase 2 completes:
-```bash
-python scripts/brand_scanner.py --brand "Client Name" --domain example.com
-```
-
-Or trigger it via: `/geo brands`
-
-The Brand Authority score is then incorporated into the final GEO Score calculation.

package/assets/worker-kits/growthub-geo-seo-v1/examples/citability-sample.md

@@ -1,155 +0,0 @@
-# Citability Analysis — Sample (GrowthHub.com)
-
-> Example: `examples/citability-sample.md`
-> This is a filled sample based on a fictitious citability analysis of thegrowthub.com.
-> Use this as a reference for what a completed CitabilityAnalysis looks like in production.
-
----
-
-## URL Audited
-
-| Field | Value |
-|---|---|
-| Target URL | https://thegrowthub.com |
-| Client | Growthub |
-| Analysis Date | 2026-04-14 |
-| Execution Mode | local-fork |
-| Script Used | scripts/citability_scorer.py |
-
----
-
-## Citability Score
-
-| Metric | Value |
-|---|---|
-| Raw Score | **71 / 100** |
-| Letter Grade | **B** |
-| Percentile Estimate | Top 33% of audited pages |
-| Score Date | 2026-04-14 |
-| Previous Score | N/A (first audit) |
-
----
-
-## Component Breakdown
-
-| Metric | Raw Score (0–100) | Weight | Weighted Score | Status |
-|---|---|---|---|---|
-| Answer Block Quality | 82 | 30% | 24.6 | Strong |
-| Self-Containment | 74 | 25% | 18.5 | Adequate |
-| Structural Readability | 79 | 20% | 15.8 | Strong |
-| Statistical Density | 47 | 15% | 7.05 | Weak |
-| Uniqueness Signals | 52 | 10% | 5.2 | Adequate |
-| **TOTAL** | | **100%** | **71.15** | **B** |
-
----
-
-## Answer Block Quality Detail
-
-**Score: 82 / 100**
-
-Paragraphs evaluated: 24
-
-| Paragraph # | Subject Clear | Evidence Present | No Unresolved Pronouns | Score |
-|---|---|---|---|---|
-| 1 (Hero intro) | Yes | No | Yes | 7 |
-| 2 (Problem statement) | Yes | Yes | Yes | 9 |
-| 3 (Solution description) | Yes | Yes | No | 7 |
-| 4 (Feature 1) | Yes | Yes | Yes | 9 |
-| 5 (Feature 2) | Yes | No | No | 5 |
-| 6 (Social proof) | Yes | Yes | Yes | 10 |
-
-Average across 24 paragraphs: 8.2 / 10 (normalized to 82 / 100)
-
-**Best answerable paragraph:**
-> "GrowthHub's AI-enabled worker kits reduce campaign production time by 65% — from 12 hours of manual brief-writing and asset coordination to 4 hours of structured operator execution. Each kit ships with pre-built templates, brand kit integration, and a step-by-step workflow that any team member can follow without prior AI training."
-
-**Weakest paragraph (needs rewrite):**
-> "It integrates seamlessly with your existing tools. This means you don't have to change how you work — it just makes everything faster and more consistent for your team."
-
-*Problem: "It" and "This" in the opening sentences are unresolved. What does "it" refer to? AI systems cannot cleanly cite this paragraph.*
-
----
-
-## Self-Containment Check
-
-| Signal | Value | Status |
-|---|---|---|
-| Total word count | 1,847 words | Good (within 800–2,500 range) |
-| Pronoun count | 94 instances | |
-| Noun count | 412 instances | |
-| Pronoun-to-noun ratio | 0.23 : 1 | Good (< 0.3) |
-| Optimal word count range | 800–2,500 words | On target |
-| Contextual dependency score | 27 / 100 | Low dependency — good |
-
-**Self-containment score: 74 / 100**
-
-Primary drag: 11 paragraphs open with "This," "It," or "They" without a prior noun anchor in the same sentence. These are individually readable but drop the raw self-containment score.
-
----
-
-## Structural Readability Check
-
-| Signal | Status | Notes |
-|---|---|---|
-| H1 present | Yes | "Growthub — AI-Enabled Growth Infrastructure" |
-| H2 count | 7 | Well-structured — above the ≥2 minimum |
-| H3 count | 12 | Good depth |
-| Numbered lists | Yes | 3 numbered lists found |
-| Bulleted lists | Yes | 8 bulleted lists found |
-| Average paragraph length | 77 words | Well within the ≤150 word target |
-| Wall-of-text sections | No | Longest unbroken block is 134 words (acceptable) |
-
-**Structural readability score: 79 / 100**
-
-Strong performance. Minor deduction: 2 of the 7 H2 sections have only a single H3 child — slight structural thinness.
-
----
-
-## Statistical Density Check
-
-| Signal | Value | Status |
-|---|---|---|
-| Percentage references | 4 instances | Low |
-| Numbered data points | 6 instances | Low |
-| Dollar/currency figures | 0 instances | Missing |
-| Year/date references | 3 instances | |
-| Data points per 1,000 words | 7.0 | Slightly below optimal (8–15) |
-| Statistical density score | 47 / 100 | Weak |
-
-**Key finding:** Statistical density is the weakest component. The homepage copy is benefit-focused but lacks specific performance numbers, client results, or category statistics. Adding 6–8 concrete data points (e.g., "reduces production time by 65%," "used by 120+ agencies") would push this score above 70.
-
----
-
-## Uniqueness Signals Check
-
-| Signal | Present | Example |
-|---|---|---|
-| First-party study or research | No | — |
-| Proprietary data references | No | — |
-| Original methodology | Yes | "Worker kit architecture" is a proprietary concept defined on the page |
-| Unique terminology | Yes | "operator execution," "worker kit," "GEO Score" — distinctive phrasing |
-| Non-generic claims | Partial | "AI-enabled growth infrastructure" is specific but unquantified |
-
-**Uniqueness score: 52 / 100**
-
-Growthub has genuine intellectual property in its worker kit architecture. The terminology is distinctive. The score is limited by the absence of original research or first-party data claims.
-
----
-
-## Key Findings
-
-1. **Answer Block Quality is the standout strength.** The homepage copy is well-structured for AI citation — paragraphs are clear, evidence-backed, and readable. The top 40% of paragraphs score 9–10/10 individually.
-
-2. **Statistical Density is the primary drag on the citability score.** With 7.0 data points per 1,000 words (target: 8–15), the page is slightly data-thin. Adding quantified client results or performance benchmarks would be the highest-leverage single edit.
-
-3. **Self-containment is solid but improvable.** The pronoun-to-noun ratio of 0.23 is healthy, but 11 paragraphs open with unresolved pronoun references. Fixing these 11 opening sentences would add approximately 5–6 points to the self-containment score.
-
----
-
-## Top 3 Improvements
-
-| Priority | Action | Expected Score Gain | Effort |
-|---|---|---|---|
-| 1 | Add 6 quantified performance claims (%, numbers, client results) throughout the homepage | +8 to +12 points (Statistical Density) | Medium — 2–3 hours of content editing |
-| 2 | Rewrite 11 paragraph openers that begin with "It," "This," or "They" to use explicit noun subjects | +5 to +7 points (Self-Containment + Answer Block Quality) | Low — 45 minutes |
-| 3 | Publish one original data report or methodology explainer (e.g., "Worker Kit Performance Report Q1 2026") | +4 to +6 points (Uniqueness Signals) | High — 1–2 weeks of content production |