> For the complete documentation index, see [llms.txt](https://bluebricks.co/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://bluebricks.co/docs/changelog/readme.md).

# June 2026

{% updates format="full" %}
{% update date="2026-06-24" %}

## Sound on, confirmations clear

This release adds optional sound effects to the agent and surfaces tool-confirmation status right in the thread list, alongside a few fixes.

## 🚀 New features

### Sound effects for agent interactions

Subtle audio feedback for the agent chat, confirming sends, responses, successful actions, and failures. Sounds are off by default, togglable, and persist across sessions.

**What does it mean for you?**\
If you like an audible cue when the agent finishes or something needs your attention, you can turn it on. If you don't, nothing changes.

***

## ⚡ Improvements

### Pending tool confirmations shown in the thread list

The agent sidebar now visually marks threads that are waiting on a tool-confirmation approval, updating in real time.

**What does it mean for you?**\
When the agent pauses for your approval on a write action, you can spot which thread needs you without opening each one.

***

## 🐛 Bug fixes

* **Fixed saving an unchanged collection property clearing its value**: In Collections, properties, clicking Save in edit mode without changing the value wiped the field and errored on the next save attempt. Saving an unchanged property now keeps the value intact.
* **Fixed a stuck "pending confirmation" indicator**: When a tool-confirmation prompt timed out with no action, the thread's pending-confirmation indicator could stay on indefinitely. The timeout now clears it.
  {% endupdate %}

{% update date="2026-06-18" %}

## Find every thread, fresh starts everywhere

This release adds a dedicated agent threads page, refreshes the empty states across the app, and fixes a couple of settings bugs.

## 🚀 New features

### Agent threads page

A dedicated page to browse, filter, and search every agent thread you have access to. The sidebar mini-list is great for "what did I touch recently?", but it breaks down once you have more than a handful of threads. Each thread card shows its collection scope and the people involved.

**What does it mean for you?**\
Once you have a lot of agent threads, you can still find the one you need. Filter and search across all of them instead of scrolling the sidebar.

***

## ⚡ Improvements

### Refreshed empty states across the app

We updated the empty states on the main pages to give you more context on what to do next and where to find more information.

**What does it mean for you?**\
When a page has nothing in it yet, it now tells you how to get started instead of just showing a blank screen.

***

## 🐛 Bug fixes

* **Fixed collections dropdown not loading when inviting users**: In user settings, expanding the Collections dropdown in the Invite users flow didn't load the collections list. It now loads correctly.
* **Fixed delete button for properties and secrets**: The Delete option for properties and secrets, both inside a collection and in the general list, wasn't doing anything. Deletion now works.
  {% endupdate %}

{% update date="2026-06-10" %}

## Lights off, insight on

This release brings full dark mode to the app and makes the agent aware of running ingestion jobs, so you always know where your data is.

## 🚀 New features

### Full dark theme

The app now has complete dark theme support, applied consistently across the whole product. A new in-product appearance toggle lets you pick Light, System, or Dark instead of relying on browser or system settings only.

**What does it mean for you?**\
Late-night ops sessions are easier on the eyes. Pick the theme you prefer, or let it follow your system, and the whole app follows along.

***

## ⚡ Improvements

### Agent is now aware of ingestion jobs

The agent detects when infrastructure data is empty or stale because ingestion is still running, checks the ingestion job status, and reports progress proactively.

**What does it mean for you?**\
No more repeatedly asking the agent "is my data there yet?" after triggering ingestion. It tells you what's still loading and when to expect answers.

***

## 🐛 Bug fixes

* **Fixed phantom "thinking" indicator after a concluded PR flow**: After the agent concluded a PR and stopped, leaving the thread and returning re-triggered the "thinking" indicator even though nothing was running. The indicator now only shows for actual active runs.
  {% endupdate %}

{% update date="2026-06-03" %}

## Threads you can find, repos you can trust

This release brings read/unread tracking, favorites, and message chapters to agent threads, and tightens how the agent handles your GitHub repository connections.

## 🚀 New features

### Read/unread tracking for agent threads

The agent sidebar now tracks read/unread state per user. Unread threads get a visual indicator that disappears as soon as you open the thread.

**What does it mean for you?**\
You can scan the sidebar and spot which threads have new activity since you last looked, without opening each one.

***

### Favorite agent threads

Mark a thread as favorite from the thread header menu. Favorites get their own folder in the side menu.

**What does it mean for you?**\
Keep important investigations within reach. The threads you keep coming back to stay one click away.

***

### Mark message as chapter in long agent threads

Mark any agent message as a chapter to anchor long threads. A left-edge rail lets you jump between marked messages.

**What does it mean for you?**\
Long agent threads stay navigable. Drop chapter markers at key moments and jump back to them whenever you need to.

***

## ⚡ Improvements

### GitHub disconnect now propagates to agent graph access

After you disconnect GitHub, the agent stops loading repository and file context from previously-ingested graph data. The disconnect state is sent to the agent on every run.

**What does it mean for you?**\
Disconnecting GitHub now does what you'd expect: the agent stops answering with stale repo data right away, instead of quietly using cached graph content.

***

### Clearer state when the agent needs a repo you haven't connected

If some repos are already connected on your org but the one the agent is asking about isn't, the chat now shows a dedicated permissions-missing state instead of pretending the connection succeeded.

**What does it mean for you?**\
You know exactly when the agent needs access to a specific repo that isn't connected yet, instead of seeing a successful connect followed by an inexplicable failure.

***

## 🐛 Bug fixes

* **Fixed missing Enable button for Git integration on fresh orgs**: In an org with no Git integrations added yet, the Integrations → Git page didn't show an Enable button at all, so there was no way to start the flow from the UI. The button is back.
  {% endupdate %}
  {% endupdates %}