docs: align release notes with shipped search and incremental behavior (#23)

* docs: align release notes with shipped search and incremental behavior

* docs: expand changelog coverage for shipped v1.5.0 scope

Author: PatrickSys

CHANGELOG.md

@@ -2,13 +2,29 @@
 
 ## [1.4.1](https://github.com/PatrickSys/codebase-context/compare/v1.4.0...v1.4.1) (2026-01-29)
 
-
 ### Bug Fixes
 
-* **lint:** disable no-explicit-any rule for AST manipulation code ([41547da](https://github.com/PatrickSys/codebase-context/commit/41547da2aa5529dce3d539c296d5e9d79df379fe))
+- **lint:** disable no-explicit-any rule for AST manipulation code ([41547da](https://github.com/PatrickSys/codebase-context/commit/41547da2aa5529dce3d539c296d5e9d79df379fe))
 
 ## [Unreleased]
 
+### Added
+
+- **Preflight evidence lock**: `search_codebase` edit/refactor/migrate intents now return risk-aware preflight guidance with evidence lock scoring, impact candidates, preferred/avoid patterns, and related memories.
+- **Trust-aware memory handling**: Added git-aware memory pattern support and confidence decay tests so stale or malformed evidence is surfaced as lower-confidence context instead of trusted guidance.
+
+### Changed
+
+- **Search ranking**: Removed framework-specific anchor/query promotion heuristics from core ranking flow to keep retrieval behavior generic across codebases.
+- **Search transparency**: `search_codebase` now returns `searchQuality` with confidence and diagnostic signals when retrieval looks ambiguous.
+- **Incremental indexing state**: Persist indexing counters to `indexing-stats.json` and restore them on no-op incremental runs to keep status reporting accurate on large codebases.
+- **Docs**: Updated README performance section to reflect shipped incremental refresh mode (`incrementalOnly`).
+
+### Fixed
+
+- **No-op incremental stats drift**: Fixed under-reported `indexedFiles` and `totalChunks` after no-change incremental refreshes by preferring persisted stats over capped index snapshots.
+- **Memory date validation**: Invalid memory timestamps now degrade to stale evidence rather than being surfaced as semi-trusted data.
+
 ## [1.4.0] - 2026-01-28
 
 ### Added

README.md

@@ -103,6 +103,7 @@ Before an edit happens, the agent gets a preflight briefing: what to use, what t
 - Failure memories bump risk level and surface as explicit warnings
 - Confidence decay: memories age (90-day or 180-day half-life). Stale guidance gets flagged, not blindly trusted
 - Epistemic stress detection: when evidence is contradictory, stale, or too thin, the preflight card says "insufficient evidence" instead of guessing
+- Search quality transparency: `search_codebase` includes `searchQuality` (`ok`/`low_confidence`, signals, confidence, next steps) so ambiguous retrieval is explicit instead of hidden
 
 ### Discovers
 
@@ -118,13 +119,13 @@ Hybrid search (BM25 keyword 30% + vector embeddings 70%) with structured filters
 
 Tested against a real enterprise Angular codebase (~30k files):
 
-| What was measured                  | Result                                                   |
-| ---------------------------------- | -------------------------------------------------------- |
-| Internal library detection         | 336 uses of `@company/ui-toolkit` vs 3 direct PrimeNG   |
-| DI pattern consensus               | 98% `inject()` adoption detected, constructor DI flagged |
-| Test framework detection           | 74% Jest, 26% Jasmine/Karma, per-module awareness        |
-| Wrapper discovery                  | `ToastEventService`, `DialogComponent` surfaced over raw |
-| Golden file identification         | Top 5 files scoring 4-6 modern patterns each             |
+| What was measured          | Result                                                   |
+| -------------------------- | -------------------------------------------------------- |
+| Internal library detection | 336 uses of `@company/ui-toolkit` vs 3 direct PrimeNG    |
+| DI pattern consensus       | 98% `inject()` adoption detected, constructor DI flagged |
+| Test framework detection   | 74% Jest, 26% Jasmine/Karma, per-module awareness        |
+| Wrapper discovery          | `ToastEventService`, `DialogComponent` surfaced over raw |
+| Golden file identification | Top 5 files scoring 4-6 modern patterns each             |
 
 Without this context, AI agents default to generic patterns: raw PrimeNG imports, constructor injection, Jasmine syntax. With the second brain active, generated code matches the existing codebase on first attempt.
 
@@ -278,7 +279,7 @@ This tool runs locally on your machine.
 
 - **Initial indexing**: First run may take several minutes (e.g., 2-5 min for 30k files) to compute embeddings.
 - **Subsequent queries**: Instant (milliseconds) from cache.
-- **Updates**: `refresh_index` re-scans the codebase. True incremental indexing (processing only changed files) is on the roadmap.
+- **Updates**: `refresh_index` supports full or incremental mode (`incrementalOnly: true`) to process only changed files.
 
 ## Links