Threadlines v0.7.0

Highlights

Muse Code, Grok Build and Pi are now built-in agents. Claude workflows run live in the chat, and you can stop them. Agents can hand you a command to run, Cmd+F searches a chat, and replies start sooner because Codex and Claude start while you type.

• Muse Code, Grok Build and Pi are built in. All three install, update and sign in from setup or Settings › Providers. Muse Code never bills an API key unless you allow it, and you choose whether Meta may train on your chats. Grok Build runs on a SuperGrok or X Premium plan, or an xAI API key. Revert also rewinds Grok's and Pi's own memory when it can, and says so when it can't. If you added Pi as a community agent, its row offers the switch.
• Claude workflows run live in the chat. When Claude starts a workflow, or you type ultracode, a card in the chat shows its steps and agents as they work, and Stop workflow ends the whole run. The Agents tab shows each agent's transcript and the result. When Claude shares a workflow's script, you can read it there and save it as a command. Settings › Threads sets how many agents Claude aims for.
• Run the commands agents suggest. Shell commands in an agent's finished reply get a Run in terminal button, and you can send the output back to the agent. On Windows that's PowerShell commands only. Threadlines pastes risky ones, like deletes, force pushes or sudo, into the terminal for you to check instead of running them. Agents can also ask you to run a command they can't, such as a sign-in, and they get the output when it finishes.
• Find in a chat with Cmd+F. Ctrl+F, or Cmd+F on a Mac, searches the whole conversation, including the parts of a long chat that aren't on screen, and shows a count like 3 of 12. Enter and Shift+Enter move between matches, and a folded message opens to show one. Find skips work steps. Find in chat is also in the command palette, and you can change its shortcut in Settings. It's off on phones.
• Replies start sooner. Typing in a thread whose Codex or Claude agent has stopped now starts it in the background, so your message goes to an agent that's already running. You can turn that off in Settings › Threads. Messages also reach the agent faster, because the snapshot taken before each turn reuses the last one. On a big install, startup work that took about 11 seconds now takes 2, and chats open faster.

Also improved

• The browser panel opens links meant for a new tab and sign-in pop-ups. You can trust a local dev server's own HTTPS certificate, and agent screenshots fail less often.
• With two computers on a direct connection, not the relay, an agent's browser opens on the one you're using. Quitting only warns about work on this computer.
• Claude goals use the goal bar, steering reaches Claude during a long command, and its dev servers can run past 30 minutes. The Usage page says what used up your Claude limits.
• Status checks run fewer git commands. On Windows the app starts far fewer git and PowerShell processes, and a project's own git.exe never runs in place of git.
• Quitting never gets stuck. A Stop background work button ends what an agent left running, and Stop reaches Cursor, Antigravity and community agents instead of waiting 15 seconds.
• The access menu shows the level each agent really runs at, and Auto on Antigravity now accepts edits without asking. The app no longer crashes at startup for T3 Code users.

Complete technical changes

What's Changed

• fix(web): running subagents show once, beside "Agents working" by @badcuban in #462
• feat(web): find in chat with Ctrl/Cmd+F by @badcuban in #458
• fix(web): setup section headings no longer run into their boxes by @badcuban in #463
• feat: Muse Code joins as a native provider by @badcuban in #464
• fix(acp): Stop reaches ACP agents, and Cursor runs cursor-agent by @badcuban in #461
• feat(grok): Grok Build joins as a built-in provider by @badcuban in #466
• perf(server): git status checks stop waking every worktree and locking the index by @badcuban in #465
• perf(server): Windows stops starting PowerShell in a loop and freezing on it by @badcuban in #469
• perf(server): a status refresh starts half as many git processes by @badcuban in #468
• perf(server): plain git reads on Windows start one process instead of three by @badcuban in #470
• perf: faster startup, first load and chat switching by @badcuban in #472
• fix(acp): an agent's own turns open and close in order with its output by @badcuban in #474
• fix(rooms): a helper's worktree no longer moves the thread's checkout by @badcuban in #473
• feat: run suggested commands from the chat, and agents can ask you to run one by @badcuban in #467
• perf: chat snapshots send each repeated screenshot and output once by @badcuban in #476
• fix: the composer and goals follow the agent you picked when a thread's saved agent is off by @badcuban in #475
• fix(server): Windows never runs a project's own git.exe in place of git by @badcuban in #477
• fix: quitting never gets stuck, and background work can always be stopped by @badcuban in #478
• feat: Claude workflows run live in the chat, with Stop, a workflow view and the ultracode keyword by @badcuban in #471
• fix(web): the pull request bar above the composer no longer spills out when narrow by @badcuban in #479
• fix: Markdown pages show their pictures, and preview_page checks Markdown by @badcuban in #480
• fix: a running workflow's progress reads the same everywhere by @badcuban in #481
• fix(desktop): links that open a new tab now open in the browser panel by @badcuban in #482
• feat(desktop): sign-in pop-ups open as real windows that can answer the page by @badcuban in #484
• fix: agents' browser screenshots no longer fail on busy pages or a locked screen by @badcuban in #485
• fix(web): agents' screenshots of background pages no longer hang by @badcuban in #486
• feat(desktop): local dev servers with a self-made HTTPS certificate can be trusted by @badcuban in #487
• feat: Claude's workflow size moves from the reasoning menu to Settings by @badcuban in #488
• test(web): suggestion sweep test no longer flakes when it empties the composer by @badcuban in #489
• feat(web): agent pages sit on their own panel and open the answer by @badcuban in #490
• perf(server): messages reach the agent sooner by @badcuban in #491
• fix(desktop): quitting only warns about agent work on this computer by @badcuban in #492
• fix(codex): newer Codex images, partial answers and helper models no longer get dropped by @badcuban in #493
• fix(claude): dev servers Claude starts no longer stop after 30 minutes by @badcuban in #494
• feat: Claude goals in the goal bar by @badcuban in #495
• fix(web): the Agents panel keeps agents that run for a long time by @badcuban in #496
• feat(claude): tool servers can ask you to sign in or fill in a form by @badcuban in #498
• fix(claude): steering no longer waits for a running command, and agent messages never run as commands by @badcuban in #499
• fix(web): agent page documents fill their panel by @badcuban in #500
• feat(claude): the working line shows what a running command prints by @badcuban in #502
• feat(usage): Plan limits say what used your Claude limits by @badcuban in #503
• feat(server): analytics name community agents and public models by @badcuban in #497
• feat(marketing): a page for each agent, and titles search engines can use by @badcuban in #505
• feat(pi): Pi as a built-in agent by @badcuban in #504
• feat(marketing): compare pages, and agent pages that read at a glance by @badcuban in #506
• fix(composer): the access level shows what the agent actually runs at by @badcuban in #507
• fix(pull-requests): Merge leads when the host takes a branch that is behind by @badcuban in #508
• docs: refresh the README and the social image by @badcuban in #509
• fix(terminal): commands and scripts use a free tab before opening a new one by @badcuban in #510
• feat(browser): the agent's browser opens on the computer you're using by @badcuban in #511
• feat(marketing): full-app screenshots across the site, README and social images by @badcuban in #512
• test(desktop): stop a fake proxy login from tripping secret scanning by @badcuban in #514
• fix(server): a save can no longer vanish when another save is rolled back by @badcuban in #513
• fix(tooling): the Mac typecheck cache sees what tsc reads by @badcuban in #515
• ci: setup-vp follows the repo's vite-plus and pins Node 24 by @badcuban in #516
• fix(desktop): saved secrets use Electron's async safeStorage by @badcuban in #517
• chore(deps): bump voidzero-dev/setup-vp from 1.21.1 to 1.21.2 by @dependabot[bot] in #518
• chore(deps): bump the production-dependencies group with 15 updates by @dependabot[bot] in #519
• fix(desktop): T3 Code users' app no longer crashes at every startup by @badcuban in #521
• fix(server): snapshots stay fast after a file is edited again by @badcuban in #522
• chore: agents read a version-matched copy of the Effect source by @badcuban in #523
• chore(deps-dev): bump the development-dependencies group across 1 directory with 4 updates by @dependabot[bot] in #520
• chore(deps): bump @opencode/client from 2.0.30 to 2.0.31 in the production-dependencies group by @dependabot[bot] in #525
• fix(scripts): the Effect source copy no longer triggers Dependabot alerts by @badcuban in #526
• chore: Effect's likely-mistake checks run in the type check and CI by @badcuban in #527
• fix(web): the Room agent name box centers its text by @badcuban in #529
• fix(telemetry): failed turns say why instead of "unknown" by @badcuban in #528
• fix(telemetry): the readiness check names every agent by @badcuban in #530
• docs: agents no longer wait for each other's test runs by @badcuban in #531
• fix(claude): a turn that got past an API error is not failed by it later by @badcuban in #532
• feat: agents start while you type, so the first reply comes sooner by @badcuban in #533
• test(web): copied-label checks no longer race a slow click reply by @badcuban in #534
• docs(release): prepare the 0.7.0 changelog and announcement by @badcuban in #535

Full Changelog: v0.6.1...v0.7.0