chore(release): bump to 0.7.0

Async client, BM25 hybrid retrieval, per-document TTL, Semantic Kernel
connector, search_many, optimistic concurrency, weighted graph edges,
client-side rate limiting, parallel put_vectors, and a graph node
metadata-preservation fix.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: codeforstartups

CHANGELOG.md

@@ -5,16 +5,47 @@ All notable changes to dynavec are documented here. This project adheres to
 
 ## [Unreleased]
 
+## [0.7.0] - 2026-10-01
+
+Concurrency and scale: a native async client, a lexical/hybrid retrieval path,
+per-document TTL, a new framework connector, and a graph data-loss fix.
+
 ### Added
+- **Native async client** (#270) — `AsyncDynavec` offers `async`/`await` `upsert` / `search` /
+  `get` / `delete` built on non-blocking I/O, so agent and web workloads can fan out
+  concurrent calls without thread pools.
+- **BM25 lexical retriever + dense hybrid fusion** (#233, #264) — `BM25Retriever` for pure
+  lexical search and `BM25HybridRetriever` / `db.hybrid_search()` combining dense ANN and
+  sparse BM25 via Reciprocal Rank Fusion, with tunable `dense_weight` / `sparse_weight`.
+- **Per-document TTL + automatic table provisioning** (#22) — pass `ttl_seconds=` to `upsert()`
+  / `update()` for DynamoDB-native expiry; the TTL attribute is provisioned automatically.
+- **Semantic Kernel vector store connector** (#69) — `DynavecStore` / `DynavecCollection`
+  implementing Microsoft Semantic Kernel's `VectorStore` interface, including lambda-filter
+  translation to the S3 Vectors dialect.
+- **`search_many()`** (#258) — run several queries concurrently against a namespace in one call.
 - **Optimistic concurrency on `update()`** (#24) — each update stores a `version` on the
   DynamoDB item and writes with a `ConditionExpression`, so a concurrent change raises
   `ConflictError` (nothing written) instead of being silently overwritten. Pass
   `expected_version=` to guard across your own read/update cycle; `UpsertResult.version`
   returns the new version. Plain `upsert()` stays last-writer-wins and resets the version.
+- **Weighted graph edge properties** (#268) — `add_edge()` accepts `weight` and `props`,
+  stored on the edge and surfaced through graph traversal.
+- **Client-side rate limiting** (#262) — opt-in token-bucket throttling of outbound AWS calls
+  to stay under account limits.
 
 ### Changed
 - `update()` now writes DynamoDB before S3 Vectors (previously in parallel) so a conflict
   leaves both stores untouched, and reads the document with a strongly consistent `GetItem`.
+- **Parallelized `put_vectors`** (#259) — S3 Vectors writes fan out across batches with a
+  thread pool for higher ingest throughput.
+- **Strict mypy** now runs in CI (#207) and the package ships type hints throughout.
+
+### Fixed
+- **Graph node metadata preservation** (#271) — `add_edge()` / `link_docs()` no longer wipe an
+  existing node's `ntype` and `props`; unspecified attributes are preserved via
+  `if_not_exists`, while explicit values still overwrite (#272).
+- **Markdown front-matter date normalization** on ingest (#253).
+- Cross-encoder reranker type-narrowing fix (#254).
 
 ## [0.6.0] - 2026-09-25
 

README.md

@@ -503,7 +503,7 @@ db.graph_delete_node("globex", namespace="kb")
 
 ## Status
 
-**v0.6.0 (current)** — adds **query-expansion retrievers** (Multi-Query + HyDE), **learned RRF fusion weights**, **OPQ + scalar quantization**, **cross-encoder reranking**, **hybrid graph+ANN search** and graph shortest-path, **`search().explain()`**, **cache invalidation on write**, an **embedding cache**, **`S3Source`** ingestion, DynamoDB **gzip** for large text, LlamaIndex **metadata-filter translation**, an **OpenAI Assistants** tool, and CLI **namespace export/import** — on top of the v0.5 feature set and the v0.1 hybrid core.
+**v0.7.0 (current)** — adds a **native async client** (`AsyncDynavec`), a **BM25 lexical retriever + dense hybrid fusion** (`hybrid_search`), **per-document TTL** with automatic provisioning, a **Semantic Kernel** connector, concurrent **`search_many()`**, **optimistic concurrency** on `update()`, **weighted graph edges**, **client-side rate limiting**, and parallelized `put_vectors` ingest — plus a graph metadata-preservation fix — on top of the v0.6 feature set and the v0.1 hybrid core.
 
 See the full history in **[CHANGELOG.md](CHANGELOG.md)**, the browsable **[Release notes](https://codeforstartups.github.io/dynavec/docs/release-notes.html)** page, or the **[GitHub Releases](https://github.com/codeforstartups/dynavec/releases)** tab.
 

opensource/dynavec/docs/release-notes.html

@@ -85,6 +85,31 @@ <h1>Release notes</h1>
 and every version is a
 <a href="https://github.com/codeforstartups/dynavec/releases">GitHub Release</a>.</p>
 
+<h2 id="v0-7-0">0.7.0 <span class="doc__sub" style="font-weight:400">&mdash; 2026-10-01</span></h2>
+<p>Concurrency and scale: a native async client, a lexical/hybrid retrieval path, per-document TTL, a new framework connector, and a graph data-loss fix.</p>
+<h3>Added</h3>
+<ul>
+  <li><strong>Native async client</strong> &mdash; <code>AsyncDynavec</code> with <code>async</code>/<code>await</code>
+      <code>upsert</code> / <code>search</code> / <code>get</code> / <code>delete</code> for concurrent agent and web workloads.</li>
+  <li><strong>BM25 lexical + dense hybrid</strong> &mdash; <code>BM25Retriever</code>, <code>BM25HybridRetriever</code>,
+      and <code>db.hybrid_search()</code> fusing dense ANN and sparse BM25 with Reciprocal Rank Fusion.</li>
+  <li><strong>Per-document TTL</strong> &mdash; <code>ttl_seconds=</code> on <code>upsert()</code>/<code>update()</code>
+      with automatic DynamoDB TTL provisioning.</li>
+  <li><strong>Semantic Kernel connector</strong> &mdash; <code>DynavecStore</code> / <code>DynavecCollection</code>
+      implementing Microsoft Semantic Kernel's vector-store interface.</li>
+  <li><strong><code>search_many()</code></strong> &mdash; run several queries against a namespace concurrently.</li>
+  <li><strong>Optimistic concurrency</strong> on <code>update()</code> (<code>expected_version=</code>,
+      <code>ConflictError</code>); <strong>weighted graph edges</strong>; <strong>client-side rate limiting</strong>.</li>
+</ul>
+<h3>Changed</h3>
+<ul>
+  <li>Parallelized <code>put_vectors</code> across batches for higher ingest throughput; strict mypy in CI.</li>
+</ul>
+<h3>Fixed</h3>
+<ul>
+  <li><code>add_edge()</code> / <code>link_docs()</code> no longer wipe an existing node's <code>ntype</code>/<code>props</code>.</li>
+</ul>
+
 <h2 id="v0-6-0">0.6.0 <span class="doc__sub" style="font-weight:400">&mdash; 2026-09-25</span></h2>
 <p>A large release: new retrieval strategies, quantization methods, graph and cache capabilities, and more integrations.</p>
 <h3>Added</h3>

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "dynavec"
-version = "0.6.0"
+version = "0.7.0"
 description = "Serverless hybrid vector database for AWS: fuses DynamoDB (single-digit-ms metadata + document store) with Amazon S3 Vectors (billion-scale serverless ANN). A drop-in, in-your-own-account alternative to Pinecone / Qdrant / Milvus / Weaviate / OpenSearch."
 readme = "README.md"
 requires-python = ">=3.9"

src/dynavec/__init__.py

@@ -88,7 +88,7 @@
 )
 from .transforms import LambdaTransform, TransformContext, TransformPipeline
 
-__version__ = "0.6.0"
+__version__ = "0.7.0"
 
 __all__ = [
     "Dynavec",

tools/build_docs.py

@@ -845,6 +845,31 @@ def redact(ctx):
 and every version is a
 <a href="https://github.com/codeforstartups/dynavec/releases">GitHub Release</a>.</p>
 
+<h2 id="v0-7-0">0.7.0 <span class="doc__sub" style="font-weight:400">&mdash; 2026-10-01</span></h2>
+<p>Concurrency and scale: a native async client, a lexical/hybrid retrieval path, per-document TTL, a new framework connector, and a graph data-loss fix.</p>
+<h3>Added</h3>
+<ul>
+  <li><strong>Native async client</strong> &mdash; <code>AsyncDynavec</code> with <code>async</code>/<code>await</code>
+      <code>upsert</code> / <code>search</code> / <code>get</code> / <code>delete</code> for concurrent agent and web workloads.</li>
+  <li><strong>BM25 lexical + dense hybrid</strong> &mdash; <code>BM25Retriever</code>, <code>BM25HybridRetriever</code>,
+      and <code>db.hybrid_search()</code> fusing dense ANN and sparse BM25 with Reciprocal Rank Fusion.</li>
+  <li><strong>Per-document TTL</strong> &mdash; <code>ttl_seconds=</code> on <code>upsert()</code>/<code>update()</code>
+      with automatic DynamoDB TTL provisioning.</li>
+  <li><strong>Semantic Kernel connector</strong> &mdash; <code>DynavecStore</code> / <code>DynavecCollection</code>
+      implementing Microsoft Semantic Kernel's vector-store interface.</li>
+  <li><strong><code>search_many()</code></strong> &mdash; run several queries against a namespace concurrently.</li>
+  <li><strong>Optimistic concurrency</strong> on <code>update()</code> (<code>expected_version=</code>,
+      <code>ConflictError</code>); <strong>weighted graph edges</strong>; <strong>client-side rate limiting</strong>.</li>
+</ul>
+<h3>Changed</h3>
+<ul>
+  <li>Parallelized <code>put_vectors</code> across batches for higher ingest throughput; strict mypy in CI.</li>
+</ul>
+<h3>Fixed</h3>
+<ul>
+  <li><code>add_edge()</code> / <code>link_docs()</code> no longer wipe an existing node's <code>ntype</code>/<code>props</code>.</li>
+</ul>
+
 <h2 id="v0-6-0">0.6.0 <span class="doc__sub" style="font-weight:400">&mdash; 2026-09-25</span></h2>
 <p>A large release: new retrieval strategies, quantization methods, graph and cache capabilities, and more integrations.</p>
 <h3>Added</h3>