From ace25f34026db9fb7f89205603ffab010be004a3 Mon Sep 17 00:00:00 2001
From: "github-actions[bot]"
 <41898282+github-actions[bot]@users.noreply.github.com>
Date: Sat, 27 Jun 2026 12:42:28 +0000
Subject: [PATCH] =?UTF-8?q?=F0=9F=A6=8B=20New=20version=20release?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../sanitized-markdown-and-smooth-streaming.md  | 16 ----------------
 CHANGELOG.md                                    | 17 +++++++++++++++++
 package.json                                    |  2 +-
 3 files changed, 18 insertions(+), 17 deletions(-)
 delete mode 100644 .changeset/sanitized-markdown-and-smooth-streaming.md

diff --git a/.changeset/sanitized-markdown-and-smooth-streaming.md b/.changeset/sanitized-markdown-and-smooth-streaming.md
deleted file mode 100644
index 5a9ca1d..0000000
--- a/.changeset/sanitized-markdown-and-smooth-streaming.md
+++ /dev/null
@@ -1,16 +0,0 @@
----
-'@smooai/chat-widget': minor
----
-
-Render sanitized Markdown in assistant replies and citation snippets, and smooth out the streaming reveal.
-
-**Sanitized Markdown rendering.** Assistant responses and citation snippets previously showed Markdown literally (`**bold**`, numbered lists, `[links](url)` rendered as raw text). They now render to formatted HTML through a tiny, safe-by-default renderer (`markdown.ts`) that:
-
-- escapes **all** text — raw `<script>`, `<img onerror=…>`, `<iframe>` etc. render as inert text, never markup;
-- **drops images entirely** (a scraped tracking pixel can't load) — `![alt](src)` becomes its alt text;
-- allows **only `http(s)` links** (`javascript:`/`data:`/relative fall back to plain text) with `target="_blank"` + `rel="noopener noreferrer nofollow"`;
-- emits only an allowlisted tag set (`p`, `br`, `strong`, `em`, `ul`/`ol`/`li`, `code`/`pre`, `a`, `blockquote`) and **downgrades headings** to bold lines so they fit a chat bubble.
-
-User bubbles stay plain text; mid-stream text stays plain (partial Markdown renders ugly) and only the final assistant turn is rendered as Markdown. Citation snippets are also cleaned first — leading page boilerplate (logo image/link, nav, whitespace) is trimmed and the excerpt is truncated to ~260 chars at a word boundary.
-
-**Smooth streaming reveal.** The assistant bubble no longer jumps in jerky, uneven chunks as `stream_token` bursts arrive. Incoming token text is buffered and revealed via a `requestAnimationFrame` typewriter at an adaptive rate (chars-per-frame scales with the pending backlog, so it never falls behind the network); only the single streaming bubble is updated per frame (no full list rebuild), and the final turn snaps to the full Markdown render. Respects `prefers-reduced-motion` (snaps instantly) and keeps auto-scroll without fighting a visitor who has scrolled up.
diff --git a/CHANGELOG.md b/CHANGELOG.md
index fbc4960..c8ff749 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,22 @@
 # Changelog
 
+## 0.6.0
+
+### Minor Changes
+
+- 586003c: Render sanitized Markdown in assistant replies and citation snippets, and smooth out the streaming reveal.
+
+  **Sanitized Markdown rendering.** Assistant responses and citation snippets previously showed Markdown literally (`**bold**`, numbered lists, `[links](url)` rendered as raw text). They now render to formatted HTML through a tiny, safe-by-default renderer (`markdown.ts`) that:
+
+  - escapes **all** text — raw `<script>`, `<img onerror=…>`, `<iframe>` etc. render as inert text, never markup;
+  - **drops images entirely** (a scraped tracking pixel can't load) — `![alt](src)` becomes its alt text;
+  - allows **only `http(s)` links** (`javascript:`/`data:`/relative fall back to plain text) with `target="_blank"` + `rel="noopener noreferrer nofollow"`;
+  - emits only an allowlisted tag set (`p`, `br`, `strong`, `em`, `ul`/`ol`/`li`, `code`/`pre`, `a`, `blockquote`) and **downgrades headings** to bold lines so they fit a chat bubble.
+
+  User bubbles stay plain text; mid-stream text stays plain (partial Markdown renders ugly) and only the final assistant turn is rendered as Markdown. Citation snippets are also cleaned first — leading page boilerplate (logo image/link, nav, whitespace) is trimmed and the excerpt is truncated to ~260 chars at a word boundary.
+
+  **Smooth streaming reveal.** The assistant bubble no longer jumps in jerky, uneven chunks as `stream_token` bursts arrive. Incoming token text is buffered and revealed via a `requestAnimationFrame` typewriter at an adaptive rate (chars-per-frame scales with the pending backlog, so it never falls behind the network); only the single streaming bubble is updated per frame (no full list rebuild), and the final turn snaps to the full Markdown render. Respects `prefers-reduced-motion` (snaps instantly) and keeps auto-scroll without fighting a visitor who has scrolled up.
+
 ## 0.5.2
 
 ### Patch Changes
diff --git a/package.json b/package.json
index 9a74ba3..2617add 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
     "name": "@smooai/chat-widget",
-    "version": "0.5.2",
+    "version": "0.6.0",
     "description": "Embeddable AI chat as a framework-light web component — the Aurora Glass design, streaming replies, grounded sources, and per-brand theming. Speaks the smooth-operator WebSocket protocol.",
     "license": "MIT",
     "author": "SmooAI",