Padar

Changelog

Older releases — page 7 of 10.

1.4.7

Who has what running

Project settings has a new Compute tab. For each machine your project runs on it shows every workspace there — whose it is, what it reserved, and how much of that it is actually using — and what the machine has left for anybody.

New

  • DesktopProject settings has a Compute tab. Per machine: every workspace your project has there, the ticket it came from, whose it is, and how much memory it is using against what it reserved, with what the machine has left underneath. Hover a workspace to light up its share of the machine.
  • DesktopStart or stop a workspace's environment, or delete one you are finished with, without leaving the page.
  • EngineEach machine reports its own size — cores, memory and disk — so what is left on it is a measured number rather than an estimate.

Improved

  • DesktopProject settings reads down a rail on the left rather than across the top, with the project's name, repository and branch above the tabs.
1.4.6Engine

The port that looked free

When you forward a workspace to this machine, Padar warns you if something on your computer is already using one of those ports and offers to free it. Some running programs slipped past that check, so the port looked free, no warning showed, and the forward could quietly land on the wrong thing.

Fixed

  • Padar spots more of the programs already using a port on your machine. It warns you when a port you are about to forward is taken and offers to free it — but some running programs went unseen, so the port read as free and no warning appeared. Now they are caught before the forward lands on one.
1.4.5Desktop

Search that shows its work

Find in a conversation lit the whole turn a match sat in, never the word — so in a long answer you were still hunting for it. Now the word itself is highlighted, the match you’re on brightest, and searching a workspace’s files shows while it is still looking.

Improved

  • Find in a conversation (⌘F) highlights the word you searched for wherever it appears in a message, and marks the match you’re on the brightest — so you can spot it inside a long answer instead of scanning the whole turn.
  • Search across a workspace’s files (⌘⇧F) shows a spinner while it looks, so you can tell a search that is still running from one that found nothing.
1.4.4Desktop

Questions stay in the workspace that asked them

A question your agent asked in one workspace followed you into the next one, where answering it could reply to something else entirely.

Fixed

  • A question from one workspace no longer sits above the message box in another. Move away while your agent is waiting and the question stays where it was asked; come back and it is still there, still waiting for you.
1.4.3Desktop

One question at a time

Your agent can ask you several things at once, and all of them used to land in a stack between the conversation and your message box — pushing what you were reading off the screen just as you needed it. Now they come one at a time, and typing an answer counts as answering.

Improved

  • When your agent asks more than one question, you take them one at a time. Picking an option moves you to the next, the arrows move you back and forth, and the count tells you how many are left. A set of questions now takes about half the height it did, so more of the conversation stays on screen while you answer.

Fixed

  • You can answer a question by typing in the message box instead of picking one of the options. Doing that used to leave the agent waiting on a question you thought you had already answered, and the only way out was to go back and click an option you had talked past.
1.4.2Desktop

The question stays where you can answer it

When your agent asked you something, the question was drawn on the tool call that raised it — so it disappeared as soon as the conversation reloaded, with the turn still waiting on an answer you could no longer give.

Improved

  • The question sits above the conversation rather than over it, so the message that prompted it stays readable while you answer. When a turn asks more than one, you take them a card at a time and the count tells you how many are left.

Fixed

  • A question from your agent sits just above the composer and stays there until you answer it. It used to appear while the conversation was still loading and vanish once it had finished, leaving a turn stopped on a question with nothing on screen to answer.
1.4.1

Your changes, your files, and your name on them

Changes counted the branch you started from as your own work, the file tree stopped listing part-way through a large project, and a commit made in a workspace had no author at all.

Fixed

  • EngineChanges lists what you changed. On a branch with no pull request open it could show hundreds of files you never touched, and the longer the branch you started from had been moving, the more of them there were.
  • EngineThe file tree shows your whole project. On a large repo it stopped part-way through, so whole folders sat there looking empty and a file you had just added never appeared, however many times you went back to look.
  • EngineCommits made in a workspace carry your name and the GitHub account you connected. They used to be refused outright for having no author, so your agent could finish a piece of work and have no way to record it.
  • DesktopFile paths in Changes read the right way round. Long ones lost the slash between the folder and the filename and grew a stray one at the front, so every row ran together.
1.4.0Desktop

The port is the answer

A forwarded service used to say so in words, on every row, next to the number that already said it better. The word is gone and the port carries it.

Improved

  • A forwarded process no longer prints "forwarded" beside its buttons. Its port lights up instead, and clicking it copies the localhost address forwarding just gave you — whether or not the process behind it is running.
1.3.23

Decide what the agent may do before it does it

Every session used to run everything without asking, and that was the only way it ran. Now you pick — plan first, ask before each edit, or keep going as before — and when the agent needs you, you answer it on the card that asked.

New

  • DesktopChoose how a session runs, from the pill under the composer. Plan explores and proposes before writing anything. Manual asks before every edit and command. Edit automatically works inside the worktree and asks before anything wider. Auto approves what passes a safety check and pauses on anything risky. Bypass permissions is still there — it is just no longer the only option. ⇧Tab cycles them without leaving the composer, and the mode you pick sticks to that conversation.
  • DesktopWhen the agent needs permission, you answer on the tool card that asked for it, rather than hunting for a separate prompt. It can also stop and ask you a real question mid-run, and wait for your pick before it carries on.
  • MobilePermissions and questions reach the phone too. Answer them from a sheet, and change the mode of a running session without going back to the desktop.
  • DesktopA run in a workspace you made tells you when it finishes or fails, with nothing to switch on first. It stays quiet only while you are actually looking at that workspace — an app left open on a long run was the one case guaranteed to tell you nothing. Switch it off per device in Settings under Account.

Improved

  • DesktopThe model menu says when it is offering built-in defaults instead of the models your account can actually use, and it goes and fetches the real list when you open a workspace. It used to stop trying shortly after the app started, and never say which list you were looking at.

Fixed

  • EngineSigning a workspace in takes effect straight away. It could carry on using the previous credential indefinitely, with nothing to tell you it had.
  • EngineOne workspace failing to start a session no longer stops every other workspace starting one.
1.3.22

How long has it actually been working?

The count beside a working agent belongs to the turn now, not to how long you have been watching it. Leave the workspace, come back, and it is still counting.

New

  • EngineThe agent can put a screenshot on a Linear ticket. It uploads the picture and hands back a link that renders in a comment or in the ticket description.

Fixed

  • DesktopThe timer beside a working agent keeps counting when you leave a workspace and come back. It used to start again from zero, so you could not tell a turn that had been running ten seconds from one that had been running ten minutes — which is usually why you left it in the first place.

Three hours of work.Two days to ship.

Agents made writing code fast. Deciding is what’s left.Padar puts the whole team in the same workspace. Upgrade your collaboration now.

Also on
Or talk to us