v2.1.1-beta.2

AI Code Conductor Beta v2.1.1-beta.2

Beta Release (pre-release)

Codex becomes a full second assistant beside Claude Code. One Accounts page covers both: turn each on or off, see whether it is installed and up to date, and keep more than one Codex account, each with its own sign-in. A saved config picks the Codex account it runs under, a Codex session can restart into an earlier conversation, and each assistant can ask the other to review its work. A new install asks which assistants you use, an update asks once whether you use Codex, and the app can run with Codex alone. The Usage page, the usage strip and Tokenomics cover both assistants, with each account's allowance and spend shown apart. A Codex session now works as a Claude session does: it keeps its conversation through a restart, moves to another Codex account with Switch Account, shows in Logs, and has the attention dot, the Session Watchdog, Plan mode and Compact. Sentinel watches Codex too. The Agent Canvas, Vision and the in-app browser, Ask Conductor, Cloud Agents and the Memory page work for Codex too, and Codex is no longer marked Beta.

Added

• Settings, Accounts is now one page for both assistants. A Providers card at the top turns Claude Code and Codex on or off, says whether each one is installed and which version, and, when Codex is missing or too old, lists the install or update commands from OpenAI's own instructions, each with Run it for me and Copy, with Check again. When the app can tell how the Codex it found was installed, it lists the update for that way (one that OpenAI's own installer put there is updated by running that installer again); when it cannot, it lists every update. Your Claude accounts and your Codex accounts are listed below it. At least one provider always stays on, and a provider cannot be switched off while anything of it is still running.
• Claude Code and Codex can both be installed from setup. Every install or update command has Run it for me, which runs it in a visible terminal once you confirm it, and Copy. The tool's own installer comes first (Anthropic's from claude.ai, OpenAI's from chatgpt.com), and its confirmation says it downloads a script from there and runs it; npm comes second, and says so when Node.js is not found. Setup checks again when the command ends, also when it fails or you press Ctrl+C on Windows, and finds the new install without restarting the app. Anthropic's installer does not put its folder on your PATH: setup then says Claude Code is installed there, and on Windows Add it to PATH for me adds that one folder to your PATH; on a Mac or Linux it shows the line to add to your shell file. Settings, Accounts and the CLI help at the foot of the window offer the same. The setup screens where you choose the app's folders and install and set up Claude Code have Exit; the pages after them do not.
• More than one Codex account. Add one with ChatGPT or an API key; the key goes to Codex, and this app never stores it. Each Codex account signs in inside its own folder, which the app creates in its resources folder, so sessions never mix identities, and the app never reads the sign-in Codex keeps there. Each account row has a menu: Make default, Make reviewer, Sign in again, Check sign-in, Sign out, Make inactive and Archive.
• The Codex sign-in already on your computer can be used too, but only when you choose it: Use this sign-in on the Set up Codex page, or Use this computer's Codex sign-in in Settings, Accounts. The app then asks Codex whether it is signed in, and lists it only if it is. Because the app did not create it, you confirm it at each launch, and it never runs code reviews; if it is signed out, run codex login in a terminal, then Check sign-in.
• New saved config picks the Codex account a config runs under, with the default listed first; an account signed in a different way than before cannot be picked until you confirm it. A Codex session header has a Restart menu: Restart carries on with the same conversation (a new one if the session had none yet), and Restart and pick a conversation lists recent ones in the terminal to pick from, the conversations of the project's other git worktrees included, each tagged with its worktree and started in it, and led by the name you gave a session. A Codex session that was open when the app closed reopens in the same conversation, under the same account.
• Code review works both ways. A Claude session can ask for a Codex review, and a Codex session can ask for a Claude review. Each review is a separate, one-off, read-only reviewer in the project, on your reviewer account: the reviewer default, or the default account when none is set. Settings, General, Built-in Tools has a switch for each direction that names the account reviews will use and, when a review cannot run, says why. A review tool is now offered only while a review could actually run. On macOS, Claude reviews use your normal Claude sign-in.
• A fresh install asks which assistants you will use: Claude Code, Codex, or both. A computer without Claude Code can choose Use Codex only on the setup screen instead of stopping. Choosing Codex leads to a Set up Codex page that checks for the Codex CLI, lists OpenAI's installer first and then npm (and Homebrew on a Mac), each with Run it for me, which runs it in a visible terminal tab once you confirm it, and Copy, and signs you in.
• After updating, the app asks once whether you use Codex: Yes, set up Codex, or No, I don't use Codex. Codex now has accounts of its own in this app, so nothing carries over from your earlier Codex setting, and until you answer, Codex is not set up and nothing of it starts. Yes opens the Set up Codex page, where you add a Codex account or choose Use this sign-in for the Codex sign-in already on this computer; that sign-in is used only after you choose it. No keeps Codex off, and you can set it up later in Settings, Accounts.
• The Usage page covers both assistants. With Codex on, it has a Claude Code section and a Codex section. Each Codex account shows its 5-hour and weekly allowance and its plan, marked Updated when the figure comes from an open session or a fresh check, or As of, with its age, when it comes from that account's latest session. An account with no session yet says its allowance shows after the first one, and an API-key account says it is billed per token, with no plan allowance. The rail shows the page once you have two or more accounts across both assistants, and its foot links to Tokenomics for token use and cost.
• A closed Codex account signed in with ChatGPT is checked afresh when you open the Usage page, press Refresh or use a card's Retry: the app asks Codex to run its own usage check once, in that account's folder, which takes about a second. It never checks an API-key account, the sign-in Codex keeps for this computer, or an account a session or review is using, and it checks one account at a time. Closing the page stops it, and if the check fails the card shows the last reading from that account's latest session, with its time.
• The usage strip at the foot of the window shows one pill per person, grouped by provider: someone with a Claude Code account and a Codex account sees both, and percentages are never merged across providers. Settings, Status Line lists the bars to show for each provider. A limit window that has reset since the last reading says no reading since instead of showing an old figure, on the Usage page, in the strip and in a session's status line, and a Codex session's cost keeps its API eq label, with a tooltip that says API-equivalent estimate, or Estimate at API list prices for an API-key account.
• Tokenomics splits by provider and account. A Provider filter appears once you have both Claude Code and Codex usage, and an Account filter lists each account under its provider, with This computer's sign-in for the Codex folder on this computer and Not recorded for usage with no account. Each KPI shows Claude Code and Codex apart when both have usage and no filter narrows the view, and the cost chart draws a line for each when both have spend in the range; the sessions table has an Account column and each provider's mark. Codex usage is sorted by the folder it was written in; Claude usage by the account profile a local session runs under, from this version on, while SSH sessions and sessions with no account profile read Not recorded, except that a local session later resumed in the app under an account profile moves its earlier usage to that profile's account. The first start after updating sorts your earlier Codex history once, with a notice while it runs; the totals are complete throughout.
• A Codex session's card and tab now behave like a Claude session's. The attention dot lights up when Codex asks you to approve something, and a minute after a turn ends if you have not replied; the busy sweep shows while Codex works, the working pill names Codex, and the sleeping moon shows as it does for Claude. The Session Watchdog now covers Codex sessions too, under the same one switch in Settings, which stays off by default: it reads Codex's own usage-limit and server-error messages, waits out the reset, and types the retry only into Codex's empty input box. The safeguard check is Claude's only, so a Codex session's menu shows it as not available. For the attention dot, Codex asks once per Codex account to review the hooks this app gives it (Hooks need review): choose Trust all and continue. Declining leaves the session working as normal, without the dot on that account.
• The Logs page covers Codex. A local Codex session's conversation is indexed as a Claude session's is, from that session's own Codex account folder: it shows in the Logs page with its history and search, and the Logs button on a Codex tab opens it, live, with Codex's turns marked as Codex's. A Codex config has the same Index conversation logs switch. A name you give a Codex session is kept next to its conversation, so Restart and pick a conversation still shows it after the tab is closed, as the Claude list does. A Codex session's GitHub Session Context reads its own conversation. Turning Index conversation logs off, in Settings or in a config, stops indexing at once, sessions already running too, Claude and Codex alike; turning it on applies to sessions started after. What a session wrote while indexing was off is never indexed, and a conversation you resume carries on in the index where it left off, for Claude and Codex alike. Restart and pick a conversation titles a Codex conversation by its first prompt. The GitHub Session Context lists each recently edited file once, relative to the project folder, under a heading that names the session's assistant. With Codex in use, the conversation indexing notice is shown once more, as it now covers Codex's conversations too.
• A Codex config takes extra CLI arguments, in the same field a Claude Code config has, under Session startup. Each word reaches Codex as one argument. What the app sets, or what changes the account, is refused: the model, -c and the other settings flags, the permission and working-folder flags, a profile, another provider or endpoint, and a plain word such as login (or /logout) that Codex would read as one of its commands. The field says why under it and Save waits until the value is fixed (for either assistant). A Claude Code config that carries a saved value its own rule refuses now also holds Save until that value is fixed. A saved value that is refused is dropped when the session starts, and the session starts without it. Give a folder named with a plain word as the value of --add-dir: --add-dir=docs or --add-dir ./docs.
• With two or more Codex accounts, a Codex session moves to another one as a Claude session does: from the account pill at the far left of its status line, or Switch Account in its right-click menu. The session restarts on that account and carries on in the same conversation, which the app copies into that account's folder; when the conversation cannot come along whole, a note above the terminal says why and what happens instead. A Codex session's card in the sidebar shows its account too.
• A Codex session's status line shows Lines changed, counted from the edits Codex makes, and Duration, the conversation's running time carried on across restarts, as a Claude session's does. The Status Line settings apply to Codex sessions too.
• Codex models and efforts match Codex's own: the model list is the one the supported Codex versions offer in their own picker, each model offers only the effort levels it runs, and a saved effort its model cannot run is dropped. Codex prices come from the same live price list as Claude's. On a running Codex session, Compact on the status line types Codex's own /compact, and the model pill opens Codex's own model and effort picker; each types only while Codex waits at an empty prompt, and says why when it cannot.
• A Codex config can start in Plan mode, a permission choice beside Read-only, Standard, Auto and Unrestricted: the session starts read-only and the app turns on Codex's own Plan mode once Codex is ready. If it cannot, a note says so; type /plan, or /permissions to change what Codex may do. The permission pill reads plan only while Codex shows its Plan mode.
• Sentinel covers Codex. While Codex is on, it says when the installed Codex is older than the app supports or newer than it was tested with, compares the Codex model list with the one the installed Codex offers, and checks a new Codex version's release notes. Its analysis runs on the assistant you use (with both on, the one Ask Conductor runs on, Claude Code by default), on the account chosen for it in Settings, General, under Sentinel.
• Ask Conductor runs on Codex too. With Codex alone it opens on Codex, and with both assistants on, Settings, General has Ask Conductor runs on, Claude Code by default; the Ask Conductor row in the sidebar then shows which, and Sentinel's analysis follows the same choice. An open Ask tab keeps its assistant, and the choice applies the next time Ask Conductor starts. On Codex it reads the app's documentation from an AGENTS.md, answers questions about Codex itself, and can install its helper skill into the skills folder of the Codex account it runs as. Its documentation folder is rebuilt before every start, so nothing written there carries over. Codex's prompt cannot take emoji: when a question has to be typed into it, they are removed first and the sidebar row says how many, and a question that could not be typed is kept there, with Send again.
• The Agent Canvas works in Codex sessions as in Claude ones: a Codex session renders a mockup, a plan or the built site, you mark it up, and it picks up your review. A Codex account you added gets the same canvas and browser skills a Claude session gets, in its own folder; for the Codex sign-in already on this computer the app copies them into your own Codex skills folder (~/.codex/skills), marked as its own, and removes them when you turn Codex or the built-in tools off. A skill of your own with the same name, or a link of yours at that name, is never touched, and the canvas page says so, naming it; a copy of the app's that a program holds open is finished at the next pass. In a Codex session the review line is typed only once Codex's prompt is ready, and the canvas says when one could not be sent. On the Auto preset Codex refuses the canvas render and the canvas's other tools apart from the snapshot and review, because Auto cannot ask before them; use Standard or Unrestricted (see Known issues).
• Vision and the push to the in-app browser work in Codex sessions too: a Codex session can drive the vision browser, read pages as text, and send a link to its Browser pane. On the Auto preset Codex refuses these tools, because Auto cannot ask before them; use Standard or Unrestricted (see Known issues).
• Cloud Agents run on Codex too. With both assistants on, New agent asks which; a Codex agent runs on the Codex account you pick, read-only unless you tick Auto for that run, which lets it edit files in its project folder without asking (on Windows, once Codex's sandbox is set up; see Known issues).
• Insights covers Codex. Pick a Codex account in the Insights account picker, where Codex accounts are listed after Claude Code's, and New run makes that account's report. Codex has no Insights command of its own, so the app makes it: it counts the account's own Codex sessions itself, then asks Codex to write the cards, read-only and with no tools, on that account's own allowance; the ? beside the report's title says so. The report has the same cards, figures column and history as a Claude Code one, and its run is kept, so its cost shows in Tokenomics under that account. Run all makes one report over every account of both assistants; an account marked confirm at launch asks for its confirmation on a run of its own and is left out of Run all, which names it. With Claude Code off, Insights runs for Codex accounts, and a user with only Codex sessions now sees Run Insights Now instead of a message about Claude sessions.
• The Memory page lists each Codex account's own memories under the account's name, read-only (Codex keeps memories off until you turn them on with /memories), and Settings, General, Debug Logging opens each Codex account's log folder too, and shows a log_dir folder its settings name selected in the folder that holds it.
• A Codex account can sign in to chatgpt.com inside the app, as a Claude account can to claude.ai. Choose Sign in to chatgpt.com in the account's menu in Settings, Accounts, or in a Codex session's right-click menu: a sign-in window opens on that account's own browser storage, kept apart from every other account and never copied from your own browser, and closes once the sign-in and your email are confirmed. A Codex session's browser pane offers chatgpt.com on its start page, as the account the session runs under; you can sign in there too. Sign out of chatgpt.com clears it, and archiving the account clears it first; the Codex command-line tool's own Sign out leaves it alone.
• The Feature Guide has a Cloud Agents card, under Integrations, for either assistant: with Codex alone it describes Codex agents, and it says what a Codex agent's Auto tick allows (edits inside its project only, the Codex account's settings left as they are).
• Claude Opus 5.5, Claude Sonnet 5.5 and Claude Haiku 5.5 are in the model picker, and the opus, sonnet and haiku choices now mean them, as Claude Code's aliases do. A session on any of them is labelled and priced as itself, and Haiku 4.5 stays selectable as a pinned version.
• Allow Multi Spawn and Quick Start work for Codex configs as for Claude ones: a Codex config that is not Multi Spawn runs one copy at a time, and a Codex pin on Quick Start has the same copy count.
• A Codex account on paid credits shows its credits under its bars on the Usage page, in Codex credits (a count, not money): N credits, or Unlimited.
• Settings, Accounts shows Claude and Codex accounts in the same kind of row. Click an account's round chip to edit its name, colour and group, and to link it to another of your accounts, Claude or Codex, or unlink it; the name, colour and group show on every linked account, and the colour on the account's chips everywhere. A row says how many sessions are running on it, and when an account cannot be made inactive, archived or removed because sessions use it, the row names each one with Go to. Archived Codex accounts are listed under Archived, each with Restore.
• Sign in again works on a Codex account that is still signed in. The new sign-in happens in a new folder, and the account moves to it only once it is signed in, with its earlier conversations carried over, so a sign-in that fails or is cancelled leaves the account as it was. The old sign-in is kept until the app can remove it without signing out the new one: the row says so, and archiving the account removes it. The Codex sign-in already on this computer can be signed in again too, in place, after a warning that this signs out every app that uses it first.
• The title bar shows Claude's status pills only while Claude Code is on, and a Codex pill, read from OpenAI's public status page, while Codex is on.
• On Windows, a local Codex session keeps its scrollback: it runs under the console host the app ships with, so the mouse wheel scrolls back through its output as in a Claude session, and if that console host cannot start, the session starts under Windows' own instead, as do the later ones in that run; if it starts but fails at once, the session ends with nothing on screen, and starting it again uses Windows' own. Alt+V pastes an image into a Codex session: with the terminal focused, Codex attaches the image itself, and with focus elsewhere the app types a line with its path into an empty Codex prompt.
• Hello Codex: a five-page introduction to Codex in this app, shown once when you have said you use Codex and a Codex account you added is signed in. Replay it from Show the Codex introduction on the Codex card in the Feature Guide, or on the Codex row in Settings, Accounts.

Changed

• A model with no price yet reads no price in Tokenomics instead of costing $0, and a notice names it and its tokens, so the totals never quietly include a zero. A Codex model with no separate cached-input price is priced the same way in Tokenomics as in the session status line.
• The Tokenomics database stays readable by earlier versions: if you go back to one, its dashboard keeps working, and when you update again, the usage it recorded meanwhile is added back into the split views.
• A Codex session's right-click menu no longer offers Claude's account items (Open artifacts, Authenticate claude.ai and Sign in to Claude Code), which acted on your primary Claude account, and its browser pane's start page no longer offers that account's claude.ai.
• Codex is no longer marked Beta: the label is gone from setup, Settings, Accounts and the Feature Guide.
• The Feature Guide and its tour show the cards for the assistants you use. With Codex alone, the cards for features that need Claude Code (Dynamic Workflows, Multiple Accounts) and Code review are not shown, and the rest are worded for Codex; with Claude Code alone, Code review is not shown, since it needs both.
• A tip about something only one assistant has carries that assistant's mark, Claude Code's or Codex's, beside its title on the tip card and in the sidebar's tip row. Tips are not filtered by assistant, and a tip about both, or about the app itself, has no mark.
• A provider that is off starts nowhere, whichever way a session is started: a saved config says why instead of launching, and a restored or restarted tab reads Not started, names the provider and points at Settings, Accounts. The tab and its conversation are kept, so turning the provider back on and restarting the tab carries on. A tab whose launch started nothing does not count as running, so it never stops its config from launching or being deleted. The prompt to resume your last sessions names whose conversations stay resumable, and tags a session whose provider is off or not set up: it reopens as Not started, like any restored tab of that provider. With Claude Code off, Ask Conductor runs on Codex, Cloud Agents runs Codex agents only, Insights runs for Codex accounts only, and terminal-only configs still run.
• Signing a Codex account in again asks you to confirm it is the same account as before, and offers the same kind of sign-in it had. When a check finds a Codex account now signed in a different way than before, such as an API key where it had a ChatGPT sign-in, the account reads Needs attention and waits for you to confirm it with This is still my account before any session or review runs on it.
• The old Settings Codex tab and its single Codex sign-in are gone: everything they did now lives in Settings, Accounts, and Test connection is Check sign-in on each account. The ChatGPT plan label is no longer shown, because it came from a Codex sign-in file this app no longer reads. Codex sessions and Codex reviews run on this computer only in this release, and the session dialog says why its SSH options are off for Codex.
• Keyboard focus is easier to see: switches, and every field and list in Settings, show a clear ring when you reach them with Tab. Status labels on the Conductor MCP page and the badges in setup are easier to read, in both themes.
• The app now runs on Electron 44, with Chromium 152. On a Mac it needs macOS 13 (Ventura) or later, so a Mac on macOS 12 stays on an earlier version (see Known issues in the Feature Guide). From this version on, the updater does not offer a Mac a version that needs a newer macOS than the one it runs.
• Mac: this version's download has a new file name, ending in macos13.dmg, so the updater in an earlier version does not offer it. On macOS 12, stay on 2.1.1-beta.1. On macOS 13 or later, download this version's .dmg from the Releases page on GitHub once by hand; the updater offers the versions after it as before.
• A Codex review, and an Insights report, Sentinel analysis or cloud agent that runs on Codex, keeps its Codex account in use until anything it left running has been ended.
• The app starts normally even when the Insights catalogue cannot be updated or has been edited into another shape.

Fixed

• On Windows the app finds Codex as a terminal does: the first codex.exe or codex.cmd in the folders PATH lists, in their order, so a Codex installed with npm earlier in PATH is no longer passed over for a codex.exe later in it.
• On Windows the npm commands (Claude Code's install, Codex's install and update), whether the app runs them or shows them to copy, start with npm.cmd, so they work where PowerShell's script policy blocks npm.
• On Windows the app no longer freezes for a moment the first time it runs a newly installed or updated Claude Code or Codex, for example when it checks again after Add it to PATH for me, or opens the setup terminal straight after an install. That first run now happens in the background.
• Tokenomics prices Claude Sonnet 5 at Anthropic's published rate when the live price list does not have it: $2 per million input tokens and $10 per million output tokens, cache reads $0.20 and cache writes $2.50. It used $3 and $15.
• The model picker's Opus hint no longer names a context size, since Claude Code's opus alias does not mean the same size on every provider; Opus 1M still says 1M context.
• Sentinel says within a minute when its analysis cannot reach the assistant it runs on because the connection is refused (no network, or a proxy or firewall that turns it away), for Claude Code and Codex alike, instead of showing analyzing for about six minutes and then blaming a busy account or a large update. Claude Code's analysis rides out a busy service for up to eight retries and stops early when its tries get no answer; a Claude Code that prints no retry lines gives up on its own, within about two and a half minutes. A network that silently drops the traffic can still take a few minutes (see Known issues in the Feature Guide). A changelog download that answers with an error page is now reported as not read, never analysed as if it were the notes. After an analysis that failed, the Sentinel chip in the title bar says it did not complete instead of no issues found.
• Sentinel no longer runs an analysis at start when you go back to an older Claude Code version (Codex works the same way): it analyses only a version newer than the newest one it has checked, so two installs used in turn no longer start an analysis at every launch, and the panel names the version installed. Re-run in the Sentinel panel analyses the installed version and, once that analysis is done, makes it the newest one checked.
• When Close sessions could not delete the saved sessions because another program, such as a virus scanner or a sync tool, held the file, those sessions no longer come back. The app tries again at each later save, as it quits and at the next start, and offers none of them in the meantime.
• The activity chart on the Memory page keeps its dots round and its line one thickness at any window width, instead of stretching them across the card.
• SSH sessions with tmux now start Claude on a host whose login shell is zsh, the macOS default: zsh read the remote session name as a command and stopped the start line. For the same reason, End did not close that remote session or remove its files; now it does.
• Ending a session in a container that needs sudo, when the session started without the sudo password saved in the config (it was typed at the prompt), no longer reports success while Claude keeps running inside the container. End still ends the remote session and removes its files on the host, then says Claude may still be running in that container and shows a command, to run on the host, that stops Claude and removes its files in the container. To let End do it itself, save the sudo password in the config (Edit, then in Runtime enter the Sudo password with Save password left ticked); that applies to sessions started after you save it.
• End now also stops Claude in a container that has no bash (a minimal image), and closing several sessions at once (the ones selected with Ctrl-click, or Cmd-click on a Mac, or Close all sessions in a group or section) ends a container session the way closing its tab does. Before, both left Claude running inside the container.
• On Windows, an SSH session whose connection was refused straight away could close the whole app. The session now just ends, and the rest of the app keeps running.
• The Watchdog pill in a session's header now shows up, and changes, as soon as the watchdog starts or one of its checks is switched. It used to wait for some other change to the list of sessions, so a session could show no pill, or an out-of-date one.
• When the API stays overloaded, the Session Watchdog's overload retries now wait longer each time, from 30 seconds up to 5 minutes, and give up after two hours of waiting in all, for Claude Code and Codex sessions alike. Each retry that failed again used to count as a new problem, so the watchdog retried about every 30 seconds for as long as the outage lasted.
• On Windows, a Claude Code session launched from a saved config, restored, or restarted to pick a conversation now starts when Claude Code was installed with npm into a folder whose path has a space or brackets in it, such as one under a Windows user name with a space. It used to stop with an "is not recognized" error instead.
• The Session Watchdog no longer retries a Claude session after a short successful answer: it reads overload and safeguard errors in the current turn only, as it does for Codex.
• For a Claude session started in a folder with a very long path, or in a folder reached through a link, the app now finds its conversation for Logs and for reopening it, because it names that folder's transcripts folder as Claude Code does.
• Alt+V with focus outside the terminal types only the image's path into a plain terminal, quoted for the shell, and does not press Enter. On macOS and Linux it types it only into an sh, bash, zsh, dash or ksh shell; with any other shell there, or over SSH, it types nothing and says where the image was saved. In a session's partner view it types the path into the partner shell, never into the assistant behind it. On Windows, Alt+V also pastes an image file you copied in File Explorer; it used to find no image.
• After the app closed unexpectedly, the next start closes the vision browser the earlier run left open, so vision can start again.
• After a Restart, the Services panel no longer shows a gap for the session, and a Restart pressed in a tab's partner view no longer counts a Not started tab as running before it has started.
• A session card and tab that need your attention are readable again in both themes, the Close button of a session's Logs pane is no longer hidden under the floating GitHub button, and the Status Line settings' Live Preview wraps instead of cutting off its last value.
• The Index conversation logs card in setup, and its tip, say what the index powers: the Logs page and the Memory page's recent sessions. Tokenomics has an index of its own, which that switch does not change.
• A new Claude conversation now appears in Logs and in search from its first message. Claude Code names a new conversation's file before it writes it, and the app gave up on the file when it was not there yet, so a new conversation, and the next one after /clear, was missing from Logs and search.
• The AI usage popover now opens above its chip in the session status strip. It used to open below the chip, where the strip cut it off, so it never showed.
• The Services panel no longer lists an ended session again, with a negative byte gap, while its tab stays open, and closing that tab logs the session's end once.
• The Usage page's Updated line now ages while the page stays open. It used to read Updated just now until the page was left or refreshed.
• A canvas review with one note now says 1 note, not 1 notes, in the line it writes into the session.
• The Cost over time chart in Tokenomics reads clearly on long ranges such as All: the dates under it are spaced to fit the width of the chart instead of running together, carry the year when the range crosses one, and its text and dots are no longer stretched or squashed with the window.
• On Windows, a new Codex session no longer starts with text such as ]10;rgb:eeee/f2f2/f7f7 already in its prompt. Codex asks the terminal for its colours as it starts and waits a tenth of a second for the answer. The terminal could answer later than that while it was still drawing, which happened more often after the move to Electron 44, and Codex then took the late answer for typed text, so the next Enter would have sent it. The app now answers Codex at once, with the colours the terminal shows.
• When the app's settings cannot be read, the built-in tools stay off until they can.
• The built-in tools' credentials are issued for each run of the app and refused once their session ends; an SSH Persistent session that stays running on its host keeps its own across a restart of the app.
• A Codex review now runs without the Codex settings file or the rules files of the account it runs on.
• Failure text from a review, a Codex Insights run, a Codex cloud agent or a Sentinel analysis that runs on Codex hides credentials written inside URLs and session values.
• On Windows a Codex session's PATH keeps only absolute folders, as it already did on macOS and Linux.
• On Windows a Claude session's PATH keeps only absolute folders too, and so does the PATH of the terminals in which the app sets up Claude Code, runs its /insights command or runs an install or update command you confirmed, and of a cloud agent, a Sentinel or Insights run, a Claude review or the check of whether one can run, the sign-in check, the Claude Code version check during onboarding, the list of versions to pin and a pinned version's install.
• On Windows the app finds node, git, Claude Code and its helper tools only in the folders PATH names or in the Windows system folder, never in the current folder.
• A Claude Code or Codex session in a network folder that its claude.cmd, claude.bat or codex.cmd launcher cannot start from is refused, with the reason (see Known issues in the Feature Guide).
• Ask Conductor's conversation list stays inside its help folder, even when the resources folder is inside a git repository or the environment names another one.
• Names and paths shown from outside the app drop more invisible characters.
• Sentinel's proposed model entries follow the same name rule as the model picker.
• A Claude Code config's extra CLI arguments are now held to a rule of the same shape as a Codex config's: an option the app sets, or one that changes the conversation, where or how the session runs, its permission mode or the settings it reads (shown in Claude Code's help or not), is refused, and so is a word Claude Code would read as a command or as a server to run the session on, or that starts or ends with a comma. A saved value the rule refuses is left off when the session starts (see Known issues in the Feature Guide).
• A Claude Code session's extra CLI arguments now come after the app's own options on its start line. On Windows each word reaches a local Claude Code session as one argument, exactly as typed.
• With Claude Code on Windows installed as claude.cmd (npm's launcher) or claude.bat, the resume list hands options to Claude Code exactly as written, or starts nothing and says why (see Known issues in the Feature Guide).
• On macOS and Linux a Claude session starts under your login shell when it is sh, bash, zsh, dash or ksh, and otherwise under zsh on macOS, or elsewhere under bash (zsh, then sh, where bash is missing); with fish or PowerShell 7.3 or later as your login shell it finds Claude Code where that shell finds it when asked to run a command, so a PATH change made only in interactive shells, such as nvm.fish's default node, is not seen (see Known issues in the Feature Guide).
• On macOS and Linux, a terminal tab whose shell is outside the sh family, such as fish or PowerShell, opens in its folder, and the app types no folder line into it.
• A session's status bar shows only its own status updates.
• The app's status line is not set up from a resources or home folder path that a shell would read as more than a path; the log says why, and a status line of your own still shows (see Known issues in the Feature Guide).
• A Claude session whose account is no longer set up here starts nothing and says so; it never runs on another account in its place.
• Closing or ending an SSH session removes the files the app last wrote for it from the host it wrote them to, when the app can reach that host with a key or a saved password, over a separate connection, Windows hosts included, also after a Restart or once the connection has ended. Nothing is typed into the session's terminal, a session left running in tmux keeps its files until End, and closing a session that never started Claude Code opens no second connection.
• The GitHub session context, Insights, the canvas review notice, saved passwords and claude.ai sign-in answer only the app's own window.
• On Windows, an account's sign-in is written only into folders the app has made readable by you alone and checked: the account's own folder, its Claude Code folder and its identity copy (see Known issues in the Feature Guide).
• A web sign-in you cancel or leave unfinished never stays signed in after a restart, for every account.
• The account view opens a link in your browser, or follows a popup, only right after your own click, tap, Enter or Space in the view.
• Signing in to chatgpt.com inside the app reaches the sign-in pages of one more sign-in method.
• Insights keeps its reports only in its own folder, and keeps Claude Code's report exactly as it was written (see Known issues in the Feature Guide).
• Insights shows no report, rather than waiting, when a report cannot be read.
• The cross-account roll-up's written analysis is shown as plain text.

Full changelog: https://github.com/nubbymong/claude-command-center/blob/v2.1.1-beta.2/CHANGELOG.md

---

Upgrading from 2.1.0-beta.5 or earlier? Download the installer from this page — that build's updater looked for the old file name.

Claude and Claude Code are Anthropic trademarks.

SHA-256 checksums are in CHECKSUMS.txt attached to this release.