0.156.0

New Features

• Choose an optional fullscreen UI with /tui for your next launch, including transcript search, mouse selection, and right-click copying. (#46732, #46734, #46883, #46895)
• Voice conversations are enabled by default, with an F8 toggle, a /voice settings picker, and bundled audio runtimes for Linux and Windows. (#44921, #46071, #44622, #44714, #44922)
• Explore account usage, token totals, and plugin and skill activity through the /usage analytics dashboard. (#45764, #45769)
• Filter tasks by status and create worktree sessions from the agent command center; worktree support is now enabled by default. (#46839, #45276, #44870)
• Customize the terminal with six new themes and view supported Mermaid diagrams and display equations directly in responses. (#46504, #46054, #45612)
• Update the local background server through /daemon, or bypass it with --no-daemon. (#45854, #46088)

Bug Fixes

• Preserve streamed answers and plans when turns fail, are interrupted, or receive subagent completion events. (#45549, #46867)
• Fix clipboard forwarding in tmux and SSH sessions and preserve tab indentation when terminals send pasted text as individual keystrokes. (#45457, #45454)
• Restore Plan mode when resuming sessions and preserve thread identity and settings when editing earlier prompts. (#45519, #45845)
• Recover login through system proxies and refresh MCP credentials when OAuth discovery returns a 503 error. (#46562, #44636)
• Prevent speech from being dropped during playback pauses or bursts of incoming audio. (#46880)
• Close sandbox isolation gaps involving inbound Windows connections, privileged Linux/macOS sockets, and writes through read-only macOS file handles. (#44639, #45984, #46500)

Documentation

• Clarify that deprecated friendly and pragmatic personality settings no longer select response styles. (#45809)
• Explain how ? matches a single character in network proxy allow and deny patterns. (#46027)

Chores

• Update bundled TLS dependencies, including OpenSSL 3.6.4 for Linux musl builds. (#45149, #45489)

Changelog

Full Changelog: rust-v0.155.0...rust-v0.156.0

• #44606 Preserve whole diagnostic attachments and report incomplete uploads @chess-oai
• #44611 Preserve root turn attribution in turn-start events @owenlin0
• #44613 Enable user verification for local Codex Desktop sessions @riley-oai
• #44615 Treat non-interactive dumb terminals as warnings in codex doctor @etraut-openai
• #44616 Simplify enterprise OAuth login helpers and expand callback tests @nicksteele-oai
• #44617 Invalidate cached Guardian approvals for unscored permission widening @jif-oai
• #44619 Allow discarded code mode tool responses to be garbage collected @jif-oai
• #44620 Support temporary and minimal filesystem grants in MXC @iceweasel-oai
• #44622 Add /voice settings to choose a voice for future conversations @bc-openai
• #44626 Add bounded environment transport for MXC launch requests @iceweasel-oai
• #44628 Expose session analytics state in Responses turn metadata @marksteinbrick-oai
• #44629 Add manual callback input to MCP OAuth login @willwang-openai
• #44631 Focus the task list when reopening the agent command center @etraut-openai
• #44636 Recover OAuth metadata discovery from 503 responses via OIDC @malsamiri-oai
• #44639 Block non-loopback inbound traffic for the Windows offline sandbox @viyatb-oai
• #44646 Honor thread analytics opt-outs when using shared clients @priyanshusingh-de
• #44650 Enforce managed model provider selection and definitions @alexsong-oai
• #44651 Keep command center errors visible and preserve drafts @etraut-openai
• #44654 Preserve missing environment variable diagnostics in Codex Doctor @etraut-openai
• #44655 Honor thread-level plugin exclusions across runtime capabilities @mzeng-openai
• #44656 Attribute turn metrics to the models used during the turn @eternal-openai
• #44658 Keep Windows sandbox private desktops alive across helper exits @chess-oai
• #44659 Preserve turn triggers across delegated agent work @joeytrasatti-openai
• #44661 Trace tool call receipt, result readiness, and code-mode dispatch @anp-oai
• #44666 Honor system reduced-motion preferences in the TUI @imac-oai
• #44669 Resolve filesystem denials with explicit path context @seanh-oai
• #44670 Restrict login setup redirects to known platform origins @hintz-openai
• #44671 Keep voice sessions alive through mute and audio backlog @bc-openai
• #44675 Refresh global instructions at model-request boundaries @vkg-oai
• #44676 Resolve permission profiles with explicit execution-host path context @seanh-oai
• #44691 Warn about ignored configuration settings @andrewgu-oai
• #44693 Preserve selected profile settings over managed new-thread defaults @alexr-oai
• #44694 Include the Windows sandbox service in release artifacts @johnl-oai
• #44701 Add a provider for thread-scoped instructions @vkg-oai
• #44711 Return to the command center after session cancellation or deletion @etraut-openai
• #44714 Bundle Linux voice runtimes and improve audio reliability @bc-openai
• #44732 Clarify folder trust prompts and add restricted widget support @etraut-openai
• #44742 Preserve editor yanks across new sessions and thread switches @bc-openai
• #44744 Make archive confirmation number shortcuts act immediately @etraut-openai
• #44746 Check folder trust after resolving the startup destination @etraut-openai
• #44747 Update quinn-proto and allow the pinned H3 Git source @richardopenai
• #44749 Preserve voice caption order when replaying TUI history @bc-openai
• #44752 Render Markdown in agent overview task details @etraut-openai
• #44755 Check folder consent before creating or resuming TUI tasks @etraut-openai
• #44814 Test approved command execution with managed unified exec disabled @jif-oai
• #44826 Expose advertised MCP server capabilities in status responses @gpeal
• #44832 Add trusted enterprise MCP auth configuration @nicksteele-oai
• #44857 Add consistent theme-based thread colors across the TUI @etraut-openai
• #44862 Preserve parent cache affinity for ephemeral forks @jif-oai
• #44865 Scope code mode callback delegates to individual executions @abhinav-oai
• #44866 Preserve originating context for yielded code-mode tool calls @abhinav-oai
• #44867 Preserve originating budgets for code mode notifications @abhinav-oai
• #44870 Enable worktrees by default and clarify local daemon errors @bc-openai
• #44872 Add managed network policy support to the Windows MXC sandbox @iceweasel-oai
• #44877 Return public key metadata from user verification enrollment @riley-oai
• #44879 Fade Astra composer stars and stabilize cursor redraws @imac-oai
• #44883 Reject token-budget history notes for unsupported starting models @jif-oai
• #44893 Expose available access programs in model discovery @faizan-oai
• #44903 Wire up the native Windows MXC helper entry point @iceweasel-oai
• #44905 Expose disabled plugin settings in the app-server API @mzeng-openai
• #44915 Remove the deprecated thread/rollback API @owenlin0
• #44921 Enable TUI voice conversations by default @bc-openai
• #44922 Bundle native voice runtimes in Windows releases @bc-openai
• #44924 Refresh the speaker format when restarting voice output @bc-openai
• #44925 Accept voice response audio before captions on quiet turns @bc-openai
• #44928 Preserve voice meter history through quiet samples @bc-openai
• #44930 Embed friendly instructions in bundled GPT-5.4 and GPT-5.5 @rhan-oai
• #44931 Stop setting YARN_NO_PROXY in the managed proxy environment @viyatb-oai
• #44932 Unify context snapshots and group requests into windows @pakrym-oai
• #44933 Remove Windows world-writable scans and warnings from the TUI @etraut-openai
• #44934 Add scenario snapshots for remote compaction and Code Mode tools @pakrym-oai
• #44935 Remove personality selection from the TUI @rhan-oai
• #44938 Add connector auth failure detection without an install URL @priyanshusingh-de
• #44939 Respect execution hosts in Windows sandbox setup @etraut-openai
• #44942 Clarify the Windows Visual C++ runtime notice for voice packages @bc-openai
• #44944 Enforce managed provider requirements on existing app-server threads @alexsong-oai
• #44945 Route TUI Windows sandbox setup through the app server @etraut-openai
• #44946 Retire Friendly and Pragmatic personality selection @rhan-oai
• #44948 Add context snapshots for async questions and plugin refresh @pakrym-oai
• #44952 Keep voice captions visible across speaker updates and history handoff @bc-openai
• #44957 Add model grouping to the agent command center @etraut-openai
• #44969 Open tasks managed elsewhere as read-only history in the command center @etraut-openai
• #44970 Show task tokens and usage estimates in the agent command center @etraut-openai
• #44976 Make context snapshot text rendering consistent @pakrym-oai
• #45035 Run DotSlash publishing directly on Ubuntu runners @charliemarsh-oai
• #45039 Use gzip compression level 6 for Codex package archives @charliemarsh-oai
• #45051 Consolidate Rust release artifact downloads @charliemarsh-oai
• #45089 Delay automatic recaps and compact their TUI layout @fcoury-oai
• #45090 Preserve conversation context and separate next actions in recaps @fcoury-oai
• #45094 Estimate history tokens from content instead of serialized envelopes @won-openai
• #45108 Cancel pending thread title generation after manual renames @fcoury-oai
• #45112 Use blueberry in the realtime background-agent test fixture @anp-oai
• #45116 Prevent multiline report notes from submitting early @etraut-openai
• #45124 Add a feature flag for asynchronous user messages @alishobeiri-oai
• #45135 Preview streaming prose before a newline arrives in the TUI @etraut-openai
• #45137 Remove Astra sparkle animation from the TUI composer @fcoury-oai
• #45149 Use OpenSSL 3.6.4 for musl builds @zm-oai
• #45169 Extract Windows sandbox setup and installation storage into the library @zm-oai
• #45176 Wire the Windows MXC sandbox into command execution @iceweasel-oai
• #45178 Split Windows sandbox cleanup into preparation and completion phases @zm-oai
• #45180 Extract shared network configuration and environment policy helpers @seanh-oai
• #45182 Validate Windows sandbox token groups before copying SIDs @zm-oai
• #45185 Bind direct tool-call metadata to invocation outputs @ningyi-oai
• #45224 Register Windows desktop uninstall ownership before sandbox setup @chess-oai
• #45248 Use captured step settings for request metadata and tool hooks @abhinav-oai
• #45255 Open new sessions directly from the command center @etraut-openai
• #45262 Route pastes into the active history search query @charliemarsh-oai
• #45271 Preserve terminal scrollback when growing the TUI viewport @fcoury-oai
• #45276 Add worktree session creation to the agents overview @etraut-openai
• #45312 Extract Windows sandbox configuration preparation into a helper @seanh-oai
• #45345 Publish opt-in provisioned macOS packages with Rust releases @riley-oai
• #45399 Cancel code mode timer tasks when cleared or the cell finishes @jif-oai
• #45409 Add session and originating window IDs to MCP request metadata @jif-oai
• #45413 Invalidate Guardian review sessions after parent history resets @jif-oai
• #45417 Extract guardian transcript selection into guardian-context @felixxia-oai
• #45418 Extract Guardian conversation bookkeeping into the reviewer crate @felixxia-oai
• #45420 Extract Guardian sampler execution into a dedicated module @felixxia-oai
• #45428 Avoid cloning MCP server status snapshot data @charliemarsh-oai
• #45439 Share tool output schemas and defer MCP envelope construction @charliemarsh-oai
• #45440 Share Apps tool catalogs without retaining unused snapshots @charliemarsh-oai
• #45441 Preserve Guardian parent response IDs across sampling requests @jif-oai
• #45445 Attribute command and plugin analytics to the invoking model @jwang-openai
• #45454 Preserve tabs in non-bracketed paste bursts @etraut-openai
• #45455 Refactor Windows sandbox setup and service helpers @zm-oai
• #45457 Fix clipboard routing for tmux and SSH sessions @fcoury-oai
• #45459 Resolve enterprise-managed MCP registrations in the catalog @nicksteele-oai
• #45461 Label rollout compression failures by stage and I/O error kind @jif-oai
• #45463 Allow dedicated listeners for managed network proxies @iceweasel-oai
• #45475 Fix fuzzy match scoring within Unicode lowercase expansions @etraut-openai
• #45487 Retain thread persistence acquisition through session cancellation @jif-oai
• #45489 Update rustls and AWS-LC dependencies in Cargo and Bazel lockfiles @zm-oai
• #45491 Remove Guardian subagent-spawner plumbing @felixxia-oai
• #45492 Split Guardian V2 async scoring into focused modules @felixxia-oai
• #45493 Make the Guardian deadline cancellation helper crate-private @felixxia-oai
• #45495 Expose effective login methods in config requirements @acrognale-oai
• #45496 Trace global user instruction loading @vkg-oai
• #45499 Send local TUI images as portable attachments to remote app servers @etraut-openai
• #45501 Render inline TeX math as Unicode in the TUI @etraut-openai
• #45502 Add managed thread lifetimes with cancellation-safe startup @jif-oai
• #45503 Add revocable network policy primitives to the HTTP client @acrognale-oai
• #45504 Allow ConPTY output to close after the last console client exits @iceweasel-oai
• #45505 Add lifecycle tracing for unified exec @anp-oai
• #45506 Allow background persistence for steered user input @bryanashley
• #45509 Share MCP tool specs until search results are selected @owenlin0
• #45513 Allow setting daybreakEnabled when starting a thread @faizan-oai
• #45515 Filter plugin-install test analytics by event type @zm-oai
• #45516 Allow configuring the Guardian prompt template @won-openai
• #45517 Use a dedicated mock server in the provider enforcement test @zm-oai
• #45518 Route Guardian reviewers through ThreadManager for inline parents @jif-oai
• #45519 Restore collaboration mode when resuming threads @etraut-openai
• #45520 Add dependencies to the Windows sandbox service @zm-oai
• #45521 Move Guardian reviewer startup into the pool @jif-oai
• #45524 Enable MXC TTY launches and managed networking in the exec server @iceweasel-oai
• #45526 Stage Python runtime wheels directly from package directories @charliemarsh-oai
• #45528 Compress larger Windows release artifacts first @charliemarsh-oai
• #45529 Expose selected workspace routing in app-server account reads @acrognale-oai
• #45533 Harden and share Windows sandbox identity helpers @zm-oai
• #45534 Honor explicit Unix socket grants in the Linux managed sandbox @viyatb-oai
• #45535 Classify tool analytics events by call origin @eddie-openai
• #45537 Move Guardian reviewer lifecycle into the extension @jif-oai
• #45542 Add service-managed package registration for Windows sandbox accounts @zm-oai
• #45543 Refactor image content to use a shared ImageReference type @kchainani-oai
• #45544 Discourage logging full image generation results @Gan-Tu
• #45546 Move daemon packages out of the standalone CLI installation @etraut-openai
• #45548 Honor prepared Unix socket permissions in Seatbelt @viyatb-oai
• #45549 Preserve streamed answers and plans when turns terminate @etraut-openai
• #45550 Add opt-in registered package execution to the Windows sandbox @zm-oai
• #45554 Use shared Bazel cache preparation in SDK CI @charliemarsh-oai
• #45556 Add attachment upload and resolution APIs and pass stores into sessions @kchainani-oai
• #45558 Seed missing daemon installs from complete local CLI packages @etraut-openai
• #45559 Resume Windows sandbox registration refresh after service restarts @zm-oai
• #45579 Copy current thread attachments into non-ephemeral forks @joeytrasatti-openai
• #45580 Add explicit daemon package replacement from the CLI @etraut-openai
• #45602 Fix retry classification for throttling and quota errors @sdcoffey
• #45612 Render standalone display math in the TUI @etraut-openai
• #45649 Add elicitation classification support to app and MCP analytics @priyanshusingh-de
• #45669 Centralize child agent configuration in the agent module @jif-oai
• #45670 Move V2 agent message delivery into AgentControl @jif-oai
• #45672 Consolidate Guardian reviewer lifecycle ownership @jif-oai
• #45676 Move spawned-agent interruption rules into AgentControl @jif-oai
• #45677 Move Guardian review reporting and denial accounting into the extension @jif-oai
• #45679 Retire the unused Guardian extension prototype API @felixxia-oai
• #45680 Restrict guardian assessment parsing and circuit breaker visibility @felixxia-oai
• #45683 Consolidate guardian transcript tests in guardian-context @felixxia-oai
• #45684 Pass ReviewModel through guardian review sessions @felixxia-oai
• #45693 Move Guardian approval routing into the reviewer extension @jif-oai
• #45711 Add startup tool allowlists for threads @jif-oai
• #45716 Classify MCP auth and approval outcomes in analytics @priyanshusingh-de
• #45729 Move Guardian reviewer configuration into the extension @jif-oai
• #45730 Separate executor sandbox selection from Windows sandbox levels @iceweasel-oai
• #45736 Route Guardian requests through /responses with identifying headers @jif-oai
• #45737 Separate Windows sandbox implementations from legacy setup modes @iceweasel-oai
• #45739 Add typed account analytics reports to the backend client @fcoury-oai
• #45740 Add account analytics data normalization to the TUI @fcoury-oai
• #45741 Add token history and credit formatting helpers for analytics @fcoury-oai
• #45742 Add account-bound authentication for analytics requests @fcoury-oai
• #45746 Explicitly gate DotSlash publishing on release success @chess-oai
• #45749 Extract reusable Bash and Zsh startup scripts @jif-oai
• #45755 Honor canonical plugin disables for shared connectors @mzeng-openai
• #45757 Wire Windows sandbox selection into managed proxy routing @iceweasel-oai
• #45760 Disable V8 optimization paths affected by array sort bugs @kliu128
• #45762 Load analytics reports with server plans and account identity checks @fcoury-oai
• #45763 Add stacked chart primitives for account analytics @fcoury-oai
• #45764 Add an account analytics dashboard to /usage @fcoury-oai
• #45765 Add Top chats to usage analytics @fcoury-oai
• #45766 Add gated plan usage history to TUI analytics @fcoury-oai
• #45768 Add consumer Top chats usage analytics @fcoury-oai
• #45769 Add an account Summary tab to Analytics @fcoury-oai
• #45770 Improve analytics chart readability and navigation @fcoury-oai
• #45772 Expose experimental analytics plan history and improve navigation @fcoury-oai
• #45779 Use native process identities for PID-managed daemons @etraut-openai
• #45780 Allow daemon updates to restore pinned packages to latest stable @etraut-openai
• #45781 Normalize bullet glyphs in the image preparation disconnect snapshot @jgershen-oai
• #45782 Preserve Guardian authorization evidence across checkpoint migration @felixxia-oai
• #45789 Preserve Guardian evidence during checkpoint migration @felixxia-oai
• #45794 Support image references by file ID in inputs and tool outputs @kchainani-oai
• #45796 Preserve ImageUserInput in the Python SDK @kchainani-oai
• #45799 Complete Windows sandbox uninstall cleanup @chess-oai
• #45805 Preserve MCP App UI metadata in tool-call events and history @victor-openai
• #45806 Restrict plugin install requests to the root thread @rennie-openai
• #45807 Record interrupted turns in managed daemon recovery snapshots @etraut-openai
• #45809 Retire the personality feature flag and document deprecated settings @rhan-oai
• #45811 Bound WSL terminal detection and handle inconclusive probes safely @etraut-openai
• #45812 Add workspace routing support for Responses requests @acrognale-oai
• #45813 Track Windows sandbox policy and per-thread executor hosts in the TUI @etraut-openai
• #45817 Add a bounded Mermaid text renderer @etraut-openai
• #45820 Continue interrupted work after managed daemon restarts @etraut-openai
• #45821 Use app-server state for TUI Windows sandbox decisions @etraut-openai
• #45822 Add opt-in response body limits to the HTTP transport @sergio-oai
• #45823 Run R2 publishing when release dependencies succeed @dkumar-oai
• #45825 Add opt-in nonfatal handling for clock read failures @vivi
• #45830 Use app-server configuration for Windows sandbox state in the TUI @etraut-openai
• #45831 Allow session-only model and reasoning selection in the TUI @etraut-openai
• #45837 Hide WSLg's duplicate root in restricted Linux sandboxes @etraut-openai
• #45845 Revert the current thread when editing an earlier TUI prompt @etraut-openai
• #45849 Preserve the app-server shutdown signal future across loop iterations @jgershen-oai
• #45852 Preserve executor path conventions in permission summaries @seanh-oai
• #45854 Add /daemon menu for local background server updates @etraut-openai
• #45863 Preserve executor path URIs in permission profile workspace roots @seanh-oai
• #45865 Reject paths in project documentation fallback filenames @etraut-openai
• #45900 Add a hidden HTTP/3 TCP tunnel command @richardopenai
• #45915 Make Code Mode wrappers transparent to Guardian model policies @jif-oai
• #45928 Bound model catalog decode errors and classify request timeouts @jif-oai
• #45933 Avoid redundant model catalog lookups in reused Guardian reviewers @jif-oai
• #45956 Record memory storage size after successful consolidation @jif-oai
• #45957 Centralize Guardian policy resolution in config and protocol @jif-oai
• #45960 Tag memory usage telemetry with the memory version @jif-oai
• #45964 Expose partial completion in rollout compression metrics @jif-oai
• #45966 Measure rollout read and materialization durations @jif-oai
• #45978 Keep selection adjacent when hiding tasks in the agents overview @etraut-openai
• #45980 Fall back to summary history for read-only conversations @etraut-openai
• #45981 Preserve session config when switching thread permission profiles @etraut-openai
• #45982 Use native DNS resolution for the network proxy on macOS @jdkula-openai
• #45983 Show a loading message when opening tasks from the agents overview @etraut-openai
• #45984 Isolate app-server Unix sockets from filesystem-restricted commands @etraut-openai
• #45985 Replace guardian review result tuples with named structs @felixxia-oai
• #45987 Centralize Guardian action preparation for review @felixxia-oai
• #46002 Extract route-aware HTTP request execution into a separate module @acrognale-oai
• #46004 Preserve attachment Unix socket grants when controller policy is omitted @open-matt
• #46006 Report clock failures again after recovery @anp-oai
• #46008 Route built-in permission selections through the app server @etraut-openai
• #46009 Centralize compaction checkpoint selection and validation @felixxia-oai
• #46010 Enable app tool result metadata with analytics controls @ningyi-oai
• #46011 Enforce managed residency when constructing API providers @andrewgu-oai
• #46013 Route permission shortcuts through the shared selection flow @etraut-openai
• #46015 Allow callers to disable executor skills per environment @vkg-oai
• #46019 Allow hosted Apps MCP contributions to override the protocol mode @teddywyly-oai
• #46020 Add an experimental rollout compression endpoint @jif-oai
• #46026 Centralize model-message resolution and rendering in codex-prompts @rhan-oai
• #46027 Document and test ? wildcards in network proxy domain patterns @viyatb-oai
• #46029 Allow browser app cleanup hooks on interrupt @syuan-oai
• #46031 Keep Noise relay streams alive after repeated handshake failures @viyatb-oai
• #46032 Require forced macOS preferences for managed configuration @joeflorencio-openai
• #46033 Preserve orchestrator skill caches across MCP runtime updates @hlevy-openai
• #46035 Add per-app tool exposure configuration @ptiet-oai
• #46036 Preserve config error causes when saving the approvals reviewer @darius-oai
• #46038 Test Windows sandbox bin DACL modification permissions @zm-oai
• #46040 Default TUI animations off when a screen reader is detected @etraut-openai
• #46041 Tighten request handling in Guardian approval tests @jgershen-oai
• #46042 Add read-only policy support to MCP tool requests @rennie-openai
• #46043 Repair expired Windows sandbox account passwords during setup @zm-oai
• #46044 Include Code Mode tool metadata in compaction prompts @ningyi-oai
• #46047 Tag rollout compression metrics by trigger @jif-oai
• #46051 Track WebSocket continuation modes and full-input send reasons @vkg-oai
• #46054 Render Mermaid code blocks as diagrams in the TUI @etraut-openai
• #46058 Attribute analytics events to realtime voice sessions @chrisdong-oai
• #46063 Trim Guardian tests and tighten request layout assertions @felixxia-oai
• #46064 Consolidate Guardian tests at shared policy and context boundaries @felixxia-oai
• #46065 Route prepared images through the attachment store @kchainani-oai
• #46066 Keep MCP user interaction on the root thread @rennie-openai
• #46067 Remove the done prefix from TUI completion timestamps @etraut-openai
• #46069 Use syntax theme colors for inline code and file paths @etraut-openai
• #46070 Suppress warnings when skill descriptions are shortened @etraut-openai
• #46071 Add a configurable F8 shortcut for voice conversations @etraut-openai
• #46072 Account for file images in context budgets and Guardian reviews @kchainani-oai
• #46073 Bound code-mode output previews across result blocks @etraut-openai
• #46075 Use captured step settings when spawning subagents @rhan-oai
• #46077 Keep the composer responsive during Command Center session creation @etraut-openai
• #46081 Mark finished empty Code Mode tool inventories as complete @nf-openai
• #46088 Add --no-daemon to bypass the shared background server @etraut-openai
• #46096 Add a one-time composer starfield for new Astra tasks @imac-oai
• #46104 Pause TUI events in the agents overview regression test @etraut-openai
• #46107 Box app-server request handler futures to reduce stack usage @etraut-openai
• #46108 Replace Sites migration state with a runtime compatibility guard @jiwon-oai
• #46112 Preserve filesystem sandbox policy context when the cwd disappears @anp-oai
• #46116 Make TUI web and image activity summaries compact and descriptive @etraut-openai
• #46117 Add opt-in automatic background server startup @etraut-openai
• #46122 Route filesystem reads and writes by their own sandbox permissions @anp-oai
• #46123 Allow model catalogs to override the V2 spawn_agent description @rhan-oai
• #46125 Fix daemon socket isolation checks for private tmp mounts @open-matt
• #46126 Record daemon startup and update telemetry with consent handling @etraut-openai
• #46179 Include sender user messages in Guardian delegation reviews @jif-oai
• #46230 Preserve configured Flex tiers without catalog or fast-mode support @sdcoffey
• #46237 Improve Windows sandbox error details and registry cleanup @chess-oai
• #46239 Prefer the provisioning service for automatic Windows sandbox setup @zm-oai
• #46241 Repair Windows sandbox access to existing runtime children @chess-oai
• #46245 Publish Guardian cached scores and coverage atomically @felixxia-oai
• #46258 Preserve uploaded image file IDs in user message display history @kchainani-oai
• #46264 Relax delegation guidance in the v2 spawn_agent description @surajs-oai
• #46266 Expand Unicode math rendering with accents, symbols, and delimiters @etraut-openai
• #46268 Add filesystem accessors bound to environment permissions @anp-oai
• #46271 Enable MXC selection through Windows sandbox configuration @iceweasel-oai
• #46278 Reduce R2 release upload concurrency and enable standard retries @anp-oai
• #46279 Preserve Guardian's reusable history prefix across approval requests @jif-oai
• #46281 Connect app-server workspace discovery to model request routing @acrognale-oai
• #46288 Add opt-in overhead timing to code-mode responses @cassirer-openai
• #46291 Filter saved reasoning overrides from requests when disabled @felixxia-oai
• #46292 Preserve selected reasoning effort for synchronous Guardian reviews @felixxia-oai
• #46293 Route skill discovery and loading through EnvironmentAccess @anp-oai
• #46294 Separate thread startup metadata from replay history @charliemarsh-oai
• #46297 Support catalog descriptions for all multi-agent V2 tools @rhan-oai
• #46300 Centralize OAuth login and refresh handling with safer diagnostics @alexsong-oai
• #46302 Validate network socket policies using the executor OS @seanh-oai
• #46303 Serialize release asset uploads to avoid secondary rate limits @imac-oai
• #46305 Avoid cloning turn items for app-server active turn lookups @charliemarsh-oai
• #46306 Preserve bio policy errors as a distinct non-retryable error @rennie-openai
• #46309 Preserve plugin caches across display metadata refreshes @vkg-oai
• #46310 Defer environment selection changes until the next turn @sayan-oai
• #46318 Add OAuth credential management for model provider gateways @alexsong-oai
• #46319 Preserve web search actions and results in exec JSON output @etraut-openai
• #46322 Set the Windows sandbox type in the pending environment test @sayan-oai
• #46323 Record active plugin inventory in turn analytics @marksteinbrick-oai
• #46324 Broaden compaction fallback to the current model @aibrahim-oai
• #46328 Avoid persisting project trust for projectless directories @viyatb-oai
• #46330 Move retry backoff into codex-async-utils @bolinfest
• #46331 Defer environment network policy validation until after composition @seanh-oai
• #46332 Dim conversation recaps in the TUI @etraut-openai
• #46333 Handle disabled Windows sandbox accounts during cleanup @chess-oai
• #46334 Share platform identity across path, network, and sandbox configuration @seanh-oai
• #46335 Keep MCP policy evaluation consistent with turn environments @sayan-oai
• #46482 Handle unknown error classifications and configure gateway OAuth @etraut-openai
• #46486 Make TUI async question replies compatible with desktop @etraut-openai
• #46487 Keep TUI exploration grouped across reasoning and nonzero exits @etraut-openai
• #46488 Split analytics tests into focused suites @rhan-oai
• #46490 Compose gateway OAuth with primary provider authentication @alexsong-oai
• #46492 Unify TUI tool output previews with a three-row limit @etraut-openai
• #46493 Allow /review during MCP startup @etraut-openai
• #46494 Keep remote workspace roots under server control @etraut-openai
• #46495 Preserve the provisioned macOS CLI's code-signing identity @riley-oai
• #46498 Allow worktree sessions to use an existing local daemon @etraut-openai
• #46499 Allow approved escalation with environment-owned network policies @sayan-oai
• #46500 Block mutating fcntls in restricted macOS Seatbelt policies @nornagon-openai
• #46501 Add configuration and feature diagnostics to report metadata @pakrym-oai
• #46503 Use catalog model display names throughout the TUI @etraut-openai
• #46504 Add six bundled TUI themes and theme-aware accents @etraut-openai
• #46505 Support catalog parameter schemas for Multi-Agent V2 tools @rhan-oai
• #46506 Share ChatGPT cookies between HTTP and WebSocket transports @jif-oai
• #46507 Run Windows sandbox tests exclusively when local @jif-oai
• #46508 Refresh the model catalog before turns after auth changes @jif-oai
• #46509 Flush completed Guardian reviews before delivering decisions @jif-oai
• #46510 Avoid cloning active turn items for metadata-only thread resumes @charliemarsh-oai
• #46511 Avoid cloning excluded turn items during thread resume @charliemarsh-oai
• #46512 Capture Guardian review checkpoints from live context @jif-oai
• #46513 Rename AgentControl to LocalAgentControl @jif-oai
• #46514 Replay guardian checkpoints into a fresh session in tests @jif-oai
• #46516 Handle completion timing in the multi-agent resume test @jif-oai
• #46517 Stabilize the TUI exit interruption test @jif-oai
• #46518 Handle delayed process startup in the Guardian network approval test @jif-oai
• #46519 Use paused time in sampler and model catalog timeout tests @jif-oai
• #46521 Use macOS member fallback in shared process-group termination helpers @jif-oai
• #46522 Enable Guardian parent-compaction reuse by default @felixxia-oai
• #46523 Default local binding to true for MXC managed networking @iceweasel-oai
• #46524 Retry busy executable launches in packaged daemon tests @felixxia-oai
• #46527 Pin WinGet publishing dependencies in the release workflow @imac-oai
• #46528 Bind executor plugin measurements to the trusted plugin version @papayo-oai
• #46529 Allow compatible feature overrides when starting the shared daemon @etraut-openai
• #46530 Gate reasoning effort updates on explicit model support @felixxia-oai
• #46531 Preserve request-level reasoning effort for memory and title workers @felixxia-oai
• #46532 Remove com.apple.runningboard from Seatbelt platform defaults @nornagon-openai
• #46533 Disable reasoning summaries by default for new TUI threads @celia-oai
• #46534 Support env_inherit in workspace_root_test @bolinfest
• #46535 Allow unrelated namespace mounts in Linux sandbox socket checks @pash-openai
• #46539 Add a command-start callback for tool lifecycle extensions @papayo-oai
• #46540 Centralize retry decisions and delays in CodexErr @anp-oai
• #46541 Add opt-in compaction after final responses @hlevy-openai
• #46542 Add authenticated remote plugin measurement references @papayo-oai
• #46543 Add bounded filesystem path diagnostics to codex doctor @etraut-openai
• #46544 Expose declared onboarding skills in plugin details @victor-openai
• #46546 Skip skill discovery when injecting items into initialized threads @hlevy-openai
• #46547 Expose a backend-independent agent control contract @jif-oai
• #46548 Advertise the control socket's WebSocket message size limit @konsti-openai
• #46551 Add a composite action to build and smoke-test Codex packages @anp-oai
• #46552 Move child completion routing into the agent controller @jif-oai
• #46553 Return LiveAgent records from agent listing @jif-oai
• #46554 Always use private desktops for legacy Windows sandboxes @iceweasel-oai
• #46555 Track executor registrations across connection refresh and recovery @viyatb-oai
• #46556 Keep step settings and approval environments consistent @sayan-oai
• #46557 Fix active-turn environment selection lookup @sayan-oai
• #46558 Add shared plugin catalog discovery APIs @TAFOYA-OAI
• #46559 Encapsulate rollout budget accounting in LocalAgentControl @jif-oai
• #46560 Fix stale environment config in the network approval test @sayan-oai
• #46561 Support explicit provider model catalog URLs @andrewgu-oai
• #46562 Add system proxy fallback for login and startup requests @abhinav-oai
• #46564 Rename plugin MCP and app extension APIs @TAFOYA-OAI
• #46565 Preserve reasoning order in TUI activity groups @etraut-openai
• #46566 Allow recovery commands when the current thread is unavailable @etraut-openai
• #46567 Separate plugin catalog listing from package resolution @TAFOYA-OAI
• #46568 Use captured environment state for permissions and daemon recovery @sayan-oai
• #46569 Set explicit turn triggers for exec and TUI requests @eddie-openai
• #46570 Tag remote model fetch duration by authentication mode @andrewgu-oai
• #46571 Preserve macOS Seatbelt exclusions in scratch directories @nornagon-openai
• #46572 Add turn-start cloud plugin discovery to the MCP extension @TAFOYA-OAI
• #46573 Add a standalone network proxy binary with JSON configuration @viyatb-oai
• #46574 Notify users when asynchronous questions arrive in the TUI @etraut-openai
• #46575 Preserve Windows package identity for sandboxed descendants @zm-oai
• #46577 Allow thread instruction providers to share updates with subagents @vinh-gpt
• #46578 Fix standalone network proxy policy initialization @etraut-openai
• #46579 Limit the agent command center to 10 recent sessions on startup @etraut-openai
• #46580 Keep Guardian reviews on the applied instruction snapshot @jif-oai
• #46583 Deny XPC service lookups in macOS Seatbelt profiles @nornagon-openai
• #46659 Move local child-process spawning into codex-utils-pty @charliemarsh-oai
• #46660 Make local child process launch settings explicit @charliemarsh-oai
• #46661 Avoid fork when spawning macOS filesystem helpers @charliemarsh-oai
• #46673 Extend server version notices to prerelease and local clients @etraut-openai
• #46680 Improve TUI contrast, keyboard hints, and picker layouts @etraut-openai
• #46691 Standardize TUI picker presentation and fix wrapping boundaries @etraut-openai
• #46692 Unify TUI picker styling and improve compact layouts @etraut-openai
• #46694 Unify TUI completion popup styling and preserve result identity @etraut-openai
• #46695 Unify TUI prompts with shared picker styling and layouts @etraut-openai
• #46697 Unify TUI picker styling and improve compact session layouts @etraut-openai
• #46708 Preserve logical source text and styles across TUI wrapping @etraut-openai
• #46709 Add compact activity rendering and preserve transcript source text @etraut-openai
• #46710 Restore rich tool details in persisted TUI transcripts @etraut-openai
• #46711 Align persisted TUI activity groups and reasoning with live output @etraut-openai
• #46712 Recover executed tool call metadata under recorder capacity pressure @ningyi-oai
• #46714 Skip polling delays in the external queue lifecycle test @charliemarsh-oai
• #46715 Split shell snapshot credential tests into focused cases @charliemarsh-oai
• #46719 Extract the TUI transcript overlay into its own module @etraut-openai
• #46720 Cache transcript layouts across measurement and rendering @etraut-openai
• #46721 Anchor transcript scrolling to entries and bound viewport rendering @etraut-openai
• #46722 Cancel pending transcript Home jumps on subsequent navigation @etraut-openai
• #46731 Render dynamic tool activity and preserve TUI history ordering @etraut-openai
• #46732 Add selection and copying to the transcript viewer @etraut-openai
• #46733 Integrate the interactive transcript into the alternate-screen TUI @etraut-openai
• #46734 Add transcript search and per-activity detail controls @etraut-openai
• #46739 Add compact transcript browsing and prompt navigation to the TUI @etraut-openai
• #46749 Stabilize transcript and composer interactions in the TUI @etraut-openai
• #46750 Preserve startup drafts and submit them when the session is ready @etraut-openai
• #46751 Add a warning footer and dedicated warnings viewer to the TUI @etraut-openai
• #46752 Add an animated Codex logo to fresh conversations and onboarding @etraut-openai
• #46832 Add independent controls for TUI visual effects @etraut-openai
• #46835 Respect reduced motion in the voice UI @etraut-openai
• #46837 Improve agent command center rows and page navigation @etraut-openai
• #46838 Refine the agent command center layout and metadata editing @etraut-openai
• #46839 Add status filter tabs to the agent command center @etraut-openai
• #46840 Simplify agent command center hints and back navigation @etraut-openai
• #46844 Select read-only permissions for temporary structured threads @etraut-openai
• #46845 Honor the system clock preference in TUI completion timestamps @etraut-openai
• #46849 Move fullscreen transcript control to TUI configuration @etraut-openai
• #46855 Increase the default terminal probe timeout from 100 ms to 250 ms @etraut-openai
• #46856 Support stadium nodes in Mermaid terminal diagrams @etraut-openai
• #46857 Extract shared text selection helpers for the TUI @etraut-openai
• #46858 Add mouse selection and editing to the fullscreen composer @etraut-openai
• #46859 Limit the welcome logo animation to onboarding @etraut-openai
• #46861 Remove analytics TUI snapshots @etraut-openai
• #46862 Refresh analytics on identity changes and compact usage menus @etraut-openai
• #46863 Stabilize usage dashboard navigation and add keyboard help @etraut-openai
• #46864 Improve usage report layouts and preserve reading positions @etraut-openai
• #46866 Enable mouse navigation in the TUI usage view @etraut-openai
• #46867 Preserve streamed answers when subagents finish @fcoury-oai
• #46877 Allow subagents to request MCP elicitation input @rennie-openai
• #46879 Count TUI launches by fullscreen transcript configuration @etraut-openai
• #46880 Preserve voice playback across pauses and packet bursts @etraut-openai
• #46882 Streamline agent command-center shortcuts and layout @etraut-openai
• #46883 Add /tui to choose the terminal UI mode for the next launch @fcoury-oai
• #46884 Enable plain clicks on transcript links and style bare URLs @etraut-openai
• #46895 Add right-click copying for transcript and composer selections @fcoury-oai
• #46897 Honor the effective terminal color level in activity charts @etraut-openai
• #46899 Make transcript list spacing uniform after streaming @fcoury-oai
• #46902 Hide Back to bottom when the current transcript tail is visible @fcoury-oai
• #46905 Identify local background servers in /status @etraut-openai
• #46910 Preserve transcript position when opening settings pickers @etraut-openai
• #46912 Keep quota warnings visible in the TUI @fcoury-oai
• #46917 Enforce current provider requirements for the model catalog @alexsong-oai
• #46922 Fix realtime V3 transcript reconciliation for handoffs @etraut-openai
• #46929 Remove the Error prefix from TUI config.toml load errors @etraut-openai
• #46931 Use shared, keymap-aware tips in the fullscreen composer @etraut-openai
• #46938 Add independent TUI rendering toggles for Mermaid, math, and tables @etraut-openai
• #46946 Populate the agents overview from background tool responses @etraut-openai
• #46952 Preserve metadata in native user verification requests @teddywyly-oai
• #46959 Add shared agent message-board extension interfaces @jif-oai
• #46962 Avoid exposing config values in codex doctor errors @jif-oai
• #46966 Add SQLite storage for local agent message boards @jif-oai
• #46978 Implement paginated queries for the local agent message board @jif-oai
• #46979 Index channel posts by timestamp for latest-message lookups @jif-oai
• #46981 Canonicalize the workspace trust key in the doctor config test @jif-oai
• #46985 Add collaboration tools for the agent message board @jif-oai
• #46989 Expand agent message board tool integration coverage @jif-oai
• #46993 Clarify Bazel lockfile verification failures @jif-oai
• #46994 Route agent delivery through captured send requests @jif-oai
• #46999 Decouple tool restrictions from session identity with ToolPolicy @felixxia-oai
• #47010 Add agent inspection without restoring unloaded runtimes @jif-oai
• #47017 Wire agent message boards into persistent multi-agent runtimes @jif-oai
• #47026 Add diagnostics for rollout compression metadata failures @jif-oai
• #47028 Align message-board tools with the multi-agent namespace @jif-oai
• #47029 Delete persisted message boards with their root threads @jif-oai
• #47030 Preserve ordered assistant context in Guardian reviews @felixxia-oai
• #47035 Defer extension tool history materialization until first access @charliemarsh-oai
• #47036 Share stored JSON payloads across code mode cells with Arc @charliemarsh-oai
• #47038 Cache OS discovery for user agents and telemetry @charliemarsh-oai
• #47039 Preserve delivered messages through post-tool hook failures @felixxia-oai
• #47042 Recover corrupt message-board storage during thread deletion @jif-oai
• #47049 Share SQLite pools across local agent message-board handles @jif-oai
• #47050 Batch message-board subscriber reads into one SQLite result @jif-oai
• #47055 Fetch local message board thread summaries in one query @jif-oai