diff --git a/notes/pdftract-3zhf.md b/notes/pdftract-3zhf.md
new file mode 100644
index 0000000..2acde71
--- /dev/null
+++ b/notes/pdftract-3zhf.md
@@ -0,0 +1,65 @@
+# Verification Note: Phase 7.2 Coordinator (pdftract-3zhf)
+
+## Work Completed
+
+### 1. Unified Detection Entry Point
+Added `TableDetector::detect(&PageContext) -> Vec<GridCandidate>` method that:
+- Runs line-based detection first (m/l/S, re/S, re/f operators)
+- Runs borderless detection second (x0 alignment heuristic)
+- Combines results from both pipelines
+
+### 2. All Child Beads Closed
+- ✅ pdftract-88sk (7.2.1): Line-based table detection
+- ✅ pdftract-3nwz (7.2.2): Borderless table detection
+- ✅ pdftract-2oqh (7.2.3): Span-to-cell assignment
+- ✅ pdftract-ilen (7.2.4): Header row detection
+- ✅ pdftract-3c4i (7.2.5): Merged cell detection
+- ✅ pdftract-5mph (7.2.6): Table JSON output schema integration
+
+### 3. Critical Tests Pass
+
+#### PASS: 5x3 Bordered Table
+- Test: `test_detect_5x3_table`
+- Location: `crates/pdftract-core/src/table/detector.rs:899`
+- Result: All 15 cells extracted with correct row/column indices
+
+#### PASS: Merged Header Cell (colspan=3)
+- Test: `test_detect_merged_cells_colspan_3_critical_test`
+- Location: `crates/pdftract-core/src/table/cell.rs:1707`
+- Result: Header cell correctly detected with colspan=3
+
+#### PASS: Borderless Table Detection
+- Test: `test_detect_borderless_3x3_table_accepted`
+- Location: `crates/pdftract-core/src/table/detector.rs:1040`
+- Result: Borderless 3-column table detected via x0 alignment
+
+#### PASS: Two-Page Table Continuation
+- Test: `test_detect_two_page_tables_basic`
+- Location: `crates/pdftract-core/src/table/output.rs:389`
+- Result: Table continuation detected and flagged correctly
+
+### 4. Schema Verification
+- File: `docs/schema/v1.0/pdftract.schema.json`
+- Table shape includes all required fields:
+  - id, bbox, rows, header_rows
+  - detection_method ("line_based" | "borderless")
+  - continued, continued_from_prev
+  - page_index
+
+### 5. Documentation
+- File: `docs/research/table-structure-reconstruction.md`
+- Contains algorithm diagrams and explanations
+
+## Test Results
+```
+cargo test --package pdftract-core --lib table
+test result: ok. 122 passed; 0 failed; 0 ignored
+```
+
+## Files Modified
+- `crates/pdftract-core/src/table/detector.rs`: Added `detect()` method
+- `crates/pdftract-core/src/table/mod.rs`: Public API exports
+
+## Commit
+- Hash: `e995808`
+- Message: `feat(pdftract-3zhf): add unified TableDetector::detect entry point`