News
Tactic Remote 1.8.2: Conversations That Survive Interruptions
1.8.0 made Tactic Remote a mobile AI coding workbench. 1.8.2 addresses the next question: that workbench has to hold up under dropped connections, killed processes, and device switches.
Follow product and engineering updates from this channel.
Browse categoryTactic Remote 1.8.2 is out.
1.8.0 moved Tactic Remote from a remote terminal tool into a mobile AI coding workbench. 1.8.2 answers the question that comes right after: a workbench like this has to survive the interruptions that are simply part of using a phone.
On mobile, interruptions are the norm. You walk into an elevator, switch to another app for a moment, lock your screen, or pick up your iPad after starting something on your iPhone. A truly reliable workbench should hand you back the full, correctly ordered conversation after any of that happens. That's what this release is about.
Conversation History Is Now Owned by the Server
Previously, chat history was assembled entirely on the client side in real time. That created a fragile point: any reconnect, app restart, or device switch could leave you with missing or out-of-order messages.
1.8.2 changes the approach. The companion server now keeps a complete record of each session and restores it automatically when you reconnect. Message ordering is determined by the server — the app no longer has to guess.
The practical difference is straightforward: history survives a dropped connection; history survives killing and relaunching the app; switching devices shows you the same conversation. Your messages and the agent's replies appear in the order they actually happened — your message will no longer show up below the reply it prompted.
Long Responses No Longer Freeze the UI
We fixed a bug where the chat UI could become completely unresponsive while rendering a very long response.
This kind of freeze is especially disruptive on mobile — the screen looks fine, but scrolling and tapping stop working. 1.8.2 rewrites how chat content is rendered, making scrolling and display smoother for long content. The longer the response, the more noticeable the improvement.
At the same time, agent responses now appear as a single complete bubble instead of being split across multiple fragments, and messages queued while the agent is running are now delivered correctly into the conversation.
Clearer Errors When Things Go Wrong
Reliability isn't just about avoiding failures — it's also about being clear when they do happen.
- Authentication failures now show a clear prompt to check your API key, rather than a generic "connection failed" message that looks like a network problem.
- Tool approval cards correctly retain their approved/denied state after a reconnect, instead of reappearing as if they'd never been acted on.
- Queued messages are delivered reliably when switching sessions, rather than being silently dropped.
A More Resilient Companion Server
This release also hardens the companion server itself.
When an unexpected internal error occurs, the server now exits cleanly and recovers quickly, rather than getting stuck in a broken state. This change applies across the Mac, Linux, and Windows server builds. For you, it means fewer situations where the server appears to be running but isn't responding.
Download
You can update Tactic Remote on the App Store.
The Mac companion app is available on the download page or via the in-app update prompt. CLI users can run npx tacticremote@latest.
Try Tactic Remote
Control your coding Agents from your phone
Connect to Claude Code, Codex, and other Agents on your Mac, Windows, or Linux computer. Check progress and send the next instruction from iPhone or iPad.