docs: Freshness-Audit 2026-07-27

CHANGELOG.md

@@ -11,7 +11,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Initial project bootstrap
 - Phase 1: Tesseract.js OCR for counter extraction
 - Phase 2: Anthropic Claude Vision integration (BYOK)
-- 9-field prioritization matrix with German recommendations
+- 9-field prioritization matrix with recommendations
 - Settings dialog for API key management
 - GitHub Pages deployment workflow
 - Try-with-sample-screenshot button for the onboarding flow

CONTRIBUTING.md

@@ -48,11 +48,11 @@ Each entry in `recommendations.json` represents one cell in the 9-field matrix a
 | `id`                | string   | Unique identifier for the matrix cell (e.g. `"quick-wins"`) |
 | `matrix_position`   | object   | `{ "impact": "high"|"medium"|"low", "effort": "low"|"medium"|"high" }` |
 | `color`             | string   | Color code: `"green"`, `"yellow"`, `"orange"`, or `"red"`  |
-| `icon`              | string   | Emoji icon displayed in the matrix cell                    |
+| `icon`              | string   | Icon name displayed in the matrix cell (e.g. `"zap"`)      |
 | `title`             | string   | English title for the matrix cell                          |
-| `priority_label`    | string   | German priority label (e.g. `"Sofort umsetzen"`)           |
-| `subtitle`          | string   | German description of the cell's meaning                   |
-| `scale_center_hint` | string   | German hint on which Scale Center section is relevant      |
+| `priority_label`    | string   | English priority label (e.g. `"Implement Now"`)             |
+| `subtitle`          | string   | English description of the cell's meaning                  |
+| `scale_center_hint` | string   | English hint on which Scale Center section is relevant     |
 | `trigger_signals`   | array    | Metrics and keywords that trigger this cell                |
 | `recommendations`   | array    | Array of recommendation objects (see below)                |
 | `contributors`      | array    | GitHub handles of contributors                             |
@@ -62,8 +62,8 @@ Each recommendation object:
 
 | Field        | Type   | Description                                                   |
 |--------------|--------|---------------------------------------------------------------|
-| `title`      | string | German title of the recommendation                            |
-| `body`       | string | German body text with actionable advice                       |
+| `title`      | string | English title of the recommendation                            |
+| `body`       | string | English body text with actionable advice                       |
 | `tags`       | array  | Technology area tags (e.g. `["Apex", "SOQL"]`)                |
 | `references` | array  | Array of `{ "url": "...", "type": "official_docs" }` objects  |
 
@@ -99,7 +99,7 @@ npm run lint
 
 ## Recommendation Content Guidelines
 
-- Recommendation bodies should be in **German** (we will add i18n later)
+- Recommendation bodies should be in **English**
 - UI labels and titles are in **English**
 - Include Salesforce documentation links where possible
 - Be specific and actionable — avoid vague advice

README.md

@@ -4,8 +4,8 @@
 
 **Diagnose your Salesforce org's performance from a single Scale Center screenshot.**
 
-[![Status](https://img.shields.io/badge/status-public%20beta-brightgreen)]()
-[![License](https://img.shields.io/badge/license-MIT-blue)]()
+![Status](https://img.shields.io/badge/status-public%20beta-brightgreen)
+![License](https://img.shields.io/badge/license-MIT-blue)
 
 </div>
 

docs/architecture.md

@@ -87,9 +87,9 @@ matrix.js (entry point)
 ### Basic Mode (OCR)
 
 1. User uploads a Scale Center screenshot (PNG/JPG)
-2. `ocr.js` crops the top 30% of the image (counter region)
-3. Tesseract.js performs text recognition in the browser
-4. Regex patterns extract six counter values
+2. `ocr.js` detects the layout, then crops each of the six counter tiles individually (per-tile OCR, see `ORG_PERF_LAYOUT`); a full-region crop with label-based regex parsing remains as a fallback for the legacy Scale Center layout
+3. Tesseract.js performs text recognition on each tile in the browser
+4. A digit-matching regex extracts the numeric value per tile
 5. `recommendations.js` maps counters to matrix cells
 6. `matrix.js` highlights relevant cells and displays the detection summary
 

src/metadata/README.md

@@ -10,7 +10,8 @@ Migrated from the orgpulse-dev prototype on 2026-04-17. Loaded only in local bui
 
 - **apexCpuAnalyzer** — SOQL/DML in loops, nested iterations (brace-aware detection)
 - **apexRowLockAnalyzer** — read-then-write patterns with context-aware confidence (batch/trigger/controller/LIMIT 1)
-- **flowAnalyzer** — RT Flow patterns: no entry filter, record ops in loops, multiple flows on same trigger, synchronous callouts
+- **flowAnalyzer** — RT Flow patterns: no entry filter, record ops in loops, subflow-in-loop, multiple flows on same trigger, synchronous callouts, Get Records inefficiencies (redundant fetches, missing filters, unindexed fields, heap-heavy fields)
+- **metadataAnalyzer** — roll-up summaries on high-contention objects, active legacy Workflow rules, workflow field updates on high-contention objects, duplicate active triggers on the same sObject
 
 ## Adding a new analyzer