synthesis(experiment B): merge merged Periscope fork into purified+PR26

docs/synthesis/purified-plus-merged-experiment-b.md

@@ -0,0 +1,39 @@
+# Synthesis experiment B — `purified+PR26` ← `merged`
+
+**Branch:** `cursor/purified-plus-merged-f559`  
+**Base for PR:** `cursor/agentsview-purified-onto-kenn-f559`  
+**Incoming:** `merged` (Periscope fork Layer 3 + ECC + context features)
+
+## Intent
+
+Canonical **upstream-modernized line** (kenn-io replay on `agentsview` mirror) remains
+the merge base. Graft Periscope fork enhancements (rename, release pipeline, context
+page, ECC, JetBrains lifecycle) onto the purified tree.
+
+## Divergence (simulated 2026-07-29)
+
+| Metric | Value |
+|--------|-------|
+| Merge-base | `6c3317ad` (agentsview mirror tip) |
+| Commits only on purified | 25 |
+| Commits only on merged | 76 |
+| Files changed (tree diff) | 2133 |
+| Simulated merge conflicts | **735** paths (symmetric with experiment A) |
+
+## Conflict hotspots (top)
+
+Same distribution as experiment A — conflict count is direction-independent;
+**resolution bias** differs.
+
+## Resolution bias (oramasys-method)
+
+1. **Upstream replay wins** for parser/sync/postgres/schema migrations.
+2. **Merged wins** for `go.mod` module path, `cmd/periscope`, branding, install scripts.
+3. **Synthesize** frontend: context components + kit-ui modernization from both sides.
+4. **Union** docs: keep purified upstream docs + merged periscope-specific guides.
+5. **Architecturally-correct:** PR #26 CI fixes (local workflow, hydrate from origin) apply to both.
+
+## Status
+
+Manifest-only commit. Full harmonization is a multi-pass integrative merge (Mode 2),
+not a single `-X ours/theirs`.