From 5311426d92b75eaed23e10cdd6ef4a63c91b76a0 Mon Sep 17 00:00:00 2001
From: techiejd <62455039+techiejd@users.noreply.github.com>
Date: Mon, 22 Jun 2026 15:03:14 +0700
Subject: [PATCH] chore: add changeset for unreleased read primitives (1.1.0)

Consolidates the unreleased changes since 1.0.1 into one minor changeset:
findByIds (#62), searchByEmbedding (#68), populateEmbedding (#70), and the
mongodb search field-parity fix (#63).

Also sets onlyUpdatePeerDependentsWhenOutOfRange so the fixed package group
resolves to a minor (1.1.0) instead of a major peer-dependency cascade.
---
 .changeset/config.json                           |  3 +++
 .../read-primitives-and-populate-embedding.md    | 16 ++++++++++++++++
 2 files changed, 19 insertions(+)
 create mode 100644 .changeset/read-primitives-and-populate-embedding.md

diff --git a/.changeset/config.json b/.changeset/config.json
index d2c86a0..30fd434 100644
--- a/.changeset/config.json
+++ b/.changeset/config.json
@@ -11,5 +11,8 @@
   "access": "public",
   "baseBranch": "main",
   "updateInternalDependencies": "patch",
+  "___experimentalUnsafeOptions_WILL_CHANGE_IN_PATCH": {
+    "onlyUpdatePeerDependentsWhenOutOfRange": true
+  },
   "ignore": []
 }
diff --git a/.changeset/read-primitives-and-populate-embedding.md b/.changeset/read-primitives-and-populate-embedding.md
new file mode 100644
index 0000000..2ca556c
--- /dev/null
+++ b/.changeset/read-primitives-and-populate-embedding.md
@@ -0,0 +1,16 @@
+---
+"payloadcms-vectorize": minor
+"@payloadcms-vectorize/pg": minor
+"@payloadcms-vectorize/cf": minor
+"@payloadcms-vectorize/mongodb": minor
+---
+
+Add vector read primitives and the ability to retrieve stored embedding vectors.
+
+- **`vectorizedPayload.findByIds({ knowledgePool, ids, populateEmbedding })`** — batch-fetch embedding records by id, returning `Record<string, EmbeddingRecord | undefined>` (misses map to `undefined`). Implemented across the pg, cf, and mongodb adapters.
+- **`vectorizedPayload.searchByEmbedding({ knowledgePool, embedding, where, limit, populateEmbedding })`** — run a vector search directly from a raw embedding vector, skipping the query-embedding step. This is the "more like this" primitive: feed it the `embedding` returned by `findByIds({ populateEmbedding: true })` to find similar content. Result shape and `where` filtering match `search()`. Unlike `search()`, it does not run reranking even on a pool configured with `rerank`, since rerankers operate on the original query text. Local API only.
+- **`populateEmbedding?: boolean`** option (default `false`) on `search()` and `searchByEmbedding()` — when `true`, each result includes its stored embedding vector. `VectorSearchResult` and `EmbeddingRecord` now expose `embedding?: number[]`.
+
+Fix:
+
+- **mongodb adapter** `search()` now returns all stored fields (including extension fields) on each result, matching the pg and cf adapters for cross-adapter parity.