Codex · Claude · six official update streams

See what changed across your AI tools.

A unified timeline for Codex and Claude apps, CLIs, and platforms, with the official English text kept intact.

6 products·78 updates in 30 days· Latest pipeline run Aug 6, 2026, 11:49 AM

Browse by product

Latest by product

Latest updates by product

RSS ↗

Codex

Codex App

View all 33 →

Codex App

Feature

Codex App 26.730.61639 component update

Aug 5, 202626.730.61639RecommendedPackage difference analysisOfficial package ↗

- Added 3 user-facing interface messages, including “Enable Full Access for Cyber model?”, “Cyber models carry a higher risk of dangerous actions. Ensure proper safeguards are in place before granting full access.”, “This comes with risks like loss or exposure of sensitive data and prompt injection. {isCyberModel, select, true {We strongly recommend selecting "Approve for me" instead, and customizing the reviewer policy for your use case.} other {You can turn this off.}} <link>Learn more</link>”. - Changed bundled component versions: browser 26.730.61309 → 26.730.61639; chrome 26.730.61309 → 26.730.61639. - Browser and Chrome automation client code is unchanged after removing embedded app-version strings. - Rebuilt or changed package components: desktop interface, Codex agent core, code-mode host, Chronicle helper, terminal native module, SQLite native module, Electron/Chromium runtime, native permissions and remote-control helpers. This identifies changed artifacts but does not infer an undocumented bug fix.

Codex App

Feature

Codex App 26.730.61309 component update

Aug 5, 202626.730.61309RecommendedPackage difference analysisOfficial package ↗

- Added 298 user-facing interface messages, including “Allow ChatGPT to use installed plugins”, “Restart {appName} to apply this change”, “Toggle plugins”, “Ask about what you were working on, get help without explaining everything again, or discover opportunities to automate repeated tasks.”, “ChatGPT can summarize your activity across the apps and websites you use without ever recording your screen or audio.”, “When turned on, ChatGPT saves text summaries of your activity across allowed apps and websites, potentially including your communications. Private web browsing and audio is never included and you can pause or clear your history and manage what’s included at any time. This feature increases token usage. <learnMore>Learn more</learnMore>”. - Changed bundled component versions: browser 26.727.51351 → 26.730.61309; chrome 26.727.51351 → 26.730.61309; record-and-replay 1.0.1000550 → 1.0.1000621; computer-use 1.0.1000550 → 1.0.1000621. - Changed Browser automation client and Chrome automation client logic. - Rebuilt or changed package components: desktop interface, Codex agent core, code-mode host, Chronicle helper, terminal native module, SQLite native module, Electron/Chromium runtime, Computer Use services, native permissions and remote-control helpers. This identifies changed artifacts but does not infer an undocumented bug fix.

Codex

Codex CLI

View all 11 →

Codex CLI

Minor

Codex CLI Release: 0.146.1

Aug 5, 20260.146.1RecommendedOfficial source ↗

## Bug Fixes - Apply safer automatic-review defaults for cyber-capable models and explain permission changes in the terminal interface. (#37057) ## Changelog Full Changelog: rust-v0.146.0...rust-v0.146.1 - #37057 [0.146] Backport safer cyber-model auto-review defaults @anp-oai

Codex CLI

Feature

Codex CLI Release: 0.146.0

Jul 29, 20260.146.0RecommendedOfficial source ↗

## New Features - Name new sessions with `/new` or `/clear`, pin important threads, and switch between side conversations without closing them. (#34605, #34840, #35011) - Support Agent Plugins manifests, workspace plugin publishing, and additional plugin marketplaces for Amazon Bedrock and Claude Code. (#35105, #35254, #34931, #34979) - Fork threads with paginated history, including temporary forks that do not appear in thread listings. (#35220, #35251) - Connect app-server to remote Code Mode hosts over WebSocket. (#35078, #35098) - Enable standalone web search for compatible custom model providers. (#34846) - Discover executor-provided skills and securely read their associated resources, including explicitly selected skills. (#35184, #35198) ## Bug Fixes - Honor configured proxies across authentication, plugin downloads, MCP authorization, remote execution, WebSockets, redirects, and LM Studio connections. (#34479, #34509, #34655, #34678, #35023, #35056, #35239) - Keep MCP connections and Apps tools current when authentication or configuration changes, reconnecting closed servers without restarting healthy connections. (#34952, #34957, #35028, #35144, #35146, #35151) - Preserve submitted messages, final responses, failed-turn errors, imported timestamps, and approval settings across interruptions, replay, imports, and forks. (#34839, #34777, #35524, #34989, #34664) - Improve terminal responsiveness and rendering, including nonblocking interrupts, keyboard handling, narrow layouts, hyperlinks, and refreshed mention results. (#35000, #35021, #34775, #34778, #35365, #35375) - Fix Windows navigation keys, reliably terminate sandboxed process trees, and preserve proxy settings during security reviews. (#34625, #34624, #35036) - Retain more available skills under tight context budgets and warn when skill catalogs must be truncated. (#34732, #34738, #34997) ## Documentation - Document shared HTTP-client usage, proxy-aware connection pooling, and safe outbound request handling. (#34669) - Clarify Windows drive-letter canonicalization for `PathUri` values. (#34667) ## Chores - Publish release artifacts, channel metadata, and installer aliases through OpenAI-hosted release infrastructure, with GitHub fallback. (#34505, #34508, #34729, #34910) - Sign and notarize bundled macOS helper executables before packaging. (#35264) - Reduce app-server serialization overhead and unnecessary request-building allocations. (#34761, #34766, #34825) - Add enterprise-plan recognition and administrator controls for in-app updates. (#35238, #35537) ## Changelog Full Changelog: rust-v0.145.0...rust-v0.146.0 - #34447 Add a route-aware HTTP client pool @copyberry - #34449 Make external session detection limits configurable @copyberry - #34451 Attribute external agent imports by provider @copyberry - #34463 Support alpha hotfix release versions @copyberry - #34469 Preserve thread settings for goal-first and forked threads @copyberry - #34476 Separate HTTP execution from request logging @copyberry - #34478 Honor `CARGO_HTTP_CAINFO` in managed proxy environments @copyberry - #34479 Re-resolve system proxy routes across redirects @copyberry - #34481 Add route-aware redirect test coverage @copyberry - #34483 Expand route-aware proxy redirect coverage @copyberry - #34490 Route backend requests through the HTTP client factory @copyberry - #34491 Route cloud environment discovery through the HTTP client pool @copyberry - #34495 Honor system proxy settings in the daemon updater @copyberry - #34497 Preserve custom arg0 for sandboxed exec-server processes @copyberry - #34505 Mirror Rust release artifacts to Cloudflare R2 @copyberry - #34506 Respect system proxies during plugin startup sync @copyberry - #34508 Publish release metadata to R2 channels @copyberry - #34509 Honor system proxy settings for remote plugins @copyberry - #34514 Add an optional releases.openai.com installer source @copyberry - #34516 Allow `numer` in codespell checks @copyberry - #34517 Pass empty inherited FDs in the Wine PTY test @copyberry - #34522 Split MCP connection manager into focused modules @copyberry - #34525 Add step-scoped data to extension contributors @copyberry - #34533 Centralize compacted rollout item construction @copyberry - #34540 Detach Git metadata commands from stdin @copyberry - #34544 Size Noise handshake buffers to their messages @copyberry - #34547 Add reciprocal rank fusion skill selection @copyberry - #34550 Test thread-scoped MCP refresh behavior @copyberry - #34551 Simplify TUI restoration for the external editor @copyberry - #34552 Remove unused RtOptions setters @copyberry - #34553 Remove the unused TUI shutdown app command @copyberry - #34558 Remove obsolete ignored tests @copyberry - #34559 Add backend client support for Codex user settings @copyberry - #34561 Extract MCP binding clients from the connection manager @copyberry - #34562 Record rollout boundaries for materialized turns @copyberry - #34563 Page through inherited thread history @copyberry - #34566 Protect fork history references during rollout cleanup @copyberry - #34570 Highlight CUDA files as C++ in the TUI @copyberry - #34573 Accept `forceRefetch` in plugin list requests @copyberry - #34578 Gate the TUI suspend restore helper on Unix @copyberry - #34581 Add routing-card lexical skill selection @copyberry - #34588 Bind MCP calls to captured catalog revisions @copyberry - #34590 Add keyed shell environment policy filters @copyberry - #34597 Enforce exact values from managed config requirements @copyberry - #34598 Skip missing paths in filesystem sandbox entries @copyberry - #34601 Sanitize skill names in injection metrics @copyberry - #34603 Allow explicitly permitted loopback proxy targets @copyberry - #34605 Allow naming sessions with `/new` and `/clear` @copyberry - #34611 Add compatibility policies for skill catalog rendering @copyberry - #34612 Detach non-interactive subprocesses from stdin @copyberry - #34613 Route Windows sandbox proxy traffic by restricting SID @copyberry - #34615 Initialize missing-path behavior in exec-server sandbox test @copyberry - #34620 Add exec-server network policy callback types @copyberry - #34621 Load paginated model context across rollout lineages @copyberry - #34622 Increase the auto-review model override test timeout @copyberry - #34624 Terminate Windows process trees with job objects @copyberry - #34625 Fix Windows TUI navigation key handling @copyberry - #34626 Scale skill metadata budgets with model context windows @copyberry - #34629 Harden Windows elevated sandbox startup @copyberry - #34630 Add a policy-aware HTTP client builder @copyberry - #34631 Migrate agent identity to the shared HTTP client @copyberry - #34636 Keep the TUI open when starting a turn fails @copyberry - #34637 Attribute review findings to repository rules @copyberry - #34640 Update Windows process-tree tests for inherited FDs @copyberry - #34641 Harden managed proxy setup for sandboxed executions @copyberry - #34643 Migrate login HTTP construction to `HttpClient` @copyberry - #34644 Verify Git plugin SHA checkouts @copyberry - #34645 Always assign response item IDs @copyberry - #34649 Propagate resolved proxy policy through auth routing @copyberry - #34650 Require auth managers to receive routing configuration @copyberry - #34651 Migrate core test support to the shared HTTP client @copyberry - #34654 Render turn diffs for foreign environment paths @copyberry - #34655 Honor configured proxy routes for auth refreshes @copyberry - #34664 Preserve approvals reviewer when forking threads @copyberry - #34667 Document `PathUri` drive letter canonicalization @copyberry - #34669 Expand codex-http-client usage guidance @copyberry - #34678 Route LM Studio requests through the shared HTTP client @copyberry - #34681 Add session headers to realtime conversation starts @copyberry - #34687 Configure Codex Auto Review model metadata @copyberry - #34708 Rename the MCP connection manager to `McpConnectionSet` @copyberry - #34713 Order unified exec lifecycle events reliably @copyberry - #34728 Skip Git enrichment for prewarm and Guardian turns @copyberry - #34729 Publish stable installer aliases to R2 @copyberry - #34732 Preserve skill catalog entries under metadata pressure @copyberry - #34733 Make MCP resource clients follow the latest runtime @copyberry - #34734 Remove step-scoped data from extension contributors @copyberry - #34738 Drop skill descriptions before omitting catalog entries @copyberry - #34744 Update skills budget tests for extension API changes @copyberry - #34746 Match core skill ordering in extension catalogs @copyberry - #34747 Register the MCP 2026-07-28 feature flag @copyberry - #34761 Reduce app-server JSON serialization overhead @copyberry - #34763 Retry websocket requests when the previous response is missing @copyberry - #34766 Reduce typed app-server request serialization overhead @copyberry - #34769 Add the git attribution extension @copyberry - #34770 Enable exec-server network policy callbacks @copyberry - #34771 Size unified mention popups to visible results @copyberry - #34772 Normalize whitespace-only lines in agent messages @copyberry - #34775 Clamp session headers to narrow terminal widths @copyberry - #34777 Include the final agent message in turn completion summaries @copyberry - #34778 Coalesce wrapped OSC 8 hyperlinks in the TUI terminal @copyberry - #34779 Use the live parent history mode when forking agents @copyberry - #34781 Upgrade Bazel Rust and LLVM dependencies @copyberry - #34784 Reject dynamic environments named `local` @copyberry - #34785 Report skill catalog truncation during rendering @copyberry - #34786 Simplify app-server integration test setup @copyberry - #34789 Avoid unnecessary post-sampling token estimates @copyberry - #34795 Remove obsolete step store from git attribution tests @copyberry - #34796 Skip syntax highlighting for lines over 4 KiB @copyberry - #34797 Suppress omission notices in core-compatible skill catalogs @copyberry - #34806 Use path URIs in shell approval keys @copyberry - #34808 Centralize SQLite connection configuration @copyberry - #34811 Fix network access rendering in sandbox prompts @copyberry - #34814 Consolidate thread startup around `StartThreadOptions` @copyberry - #34816 Support configurable realtime BEM channel prefixes @copyberry - #34819 Enable git attribution across Codex entry points @copyberry - #34823 Run code-mode tests in non-Windows Bazel CI @copyberry - #34824 Normalize Guardian review cwd reuse keys @copyberry - #34825 Reduce cloning when building Responses requests @copyberry - #34827 Remove Windows Bazel lint toolchain overrides @copyberry - #34831 Flush analytics before in-process app server shutdown @copyberry - #34835 Track compaction time in turn profiles @copyberry - #34839 Preserve user input when MCP startup is interrupted @copyberry - #34840 Add persisted thread pinning to the app server @copyberry - #34844 Remove first-party type from app metadata @copyberry - #34845 Track multi-agent mode in world state @copyberry - #34846 Allow custom providers to opt into standalone web search @copyberry - #34847 Use Guardian model limits for review sessions @copyberry - #34849 Cache remote plugin catalogs by scope @copyberry - #34850 Disable image generation for Free-plan accounts @copyberry - #34851 Use batch metadata for plugin app summaries @copyberry - #34852 Wake sleeping threads for queued agent mail @copyberry - #34877 Wait for local plugin cache refreshes in `plugin/list` @copyberry - #34883 Set a default user agent for MCP HTTP requests @copyberry - #34887 Allow disabling the multi-agent wait tool @copyberry - #34910 Prefer releases.openai.com in standalone installers @copyberry - #34930 Centralize thread MCP state in `McpRuntime` @copyberry - #34931 Use the API plugin marketplace for Amazon Bedrock @copyberry - #34940 Keep session defaults static during config batch writes @copyberry - #34952 Reuse MCP connections across runtime refreshes @copyberry - #34957 Replace closed MCP connections during reconciliation @copyberry - #34959 Handle @ in local marketplace paths @copyberry - #34962 Move MCP connection helpers into the test module @copyberry - #34969 Keep the sleep tool outside code mode @copyberry - #34978 Honor disabled redirects in route-aware HTTP clients @copyberry - #34979 Infer the bundled Claude Code plugin marketplace @copyberry - #34981 Record externally completed agent config imports @copyberry - #34986 Enforce single-writer ownership for paginated threads @copyberry - #34989 Preserve timestamps when importing external agent sessions @copyberry - #34991 Allow omitting MCP tool prefixes per server @copyberry - #34994 Honor the configured SQLite home across state consumers @copyberry - #34995 Initialize execution environments with the final HTTP policy @copyberry - #34996 Separate Codex error details from retry metadata @copyberry - #34997 Warn when skill catalogs exceed their context budget @copyberry - #35000 Make TUI turn interrupts nonblocking @copyberry - #35011 Keep side conversations open when switching threads @copyberry - #35012 Expose remote skill icon URLs through app server @copyberry - #35013 Support incremental replay of updated thread items @copyberry - #35015 Align installed app duration metrics with the legacy baseline @copyberry - #35016 Add trusted plugin script attribution @copyberry - #35020 Attribute command executions to trusted plugin scripts @copyberry - #35021 Adapt keyboard event reporting to the terminal @copyberry - #35023 Route exec-server HTTP through configured proxy policy @copyberry - #35028 Preserve refreshed Apps tools across MCP runtime updates @copyberry - #35029 Preserve plugin attribution across command approvals @copyberry - #35031 Enforce writer ownership for thread archive and deletion @copyberry - #35033 Expose Browser Use requirements through the app server @copyberry - #35034 Route environment registry requests through the shared HTTP client @copyberry - #35036 Preserve Windows sandbox proxy settings in guardian sessions @copyberry - #35048 Track app/read request duration @copyberry - #35049 Register the Guardian V2 feature flag @copyberry - #35054 Allow disabling the update_plan tool @copyberry - #35056 Route exec-server WebSockets through configured proxies @copyberry - #35059 Decouple exec-server HTTP from reqwest types @copyberry - #35063 Track deferred tool namespaces in world state @copyberry - #35065 Avoid duplicating deferred sources in tool search @copyberry - #35067 Fix Bazel test configuration for platform-specific data @copyberry - #35078 Add WebSocket transport to the code-mode host @copyberry - #35098 Support remote code-mode hosts in app-server @copyberry - #35105 Support Agent Plugins manifests @copyberry - #35106 Allow hosts to customize `wait_for_environment` descriptions @copyberry - #35144 Prewarm MCP runtime updates in the background @copyberry - #35146 Refresh MCP runtimes when session auth changes @copyberry - #35151 Reconnect MCP servers on explicit refresh @copyberry - #35164 Encapsulate MCP refresh coordination @copyberry - #35168 Route extension warnings to app-server threads @copyberry - #35172 Compact host skill paths under metadata pressure @copyberry - #35175 Wait for reloaded worker completion in the resume test @copyberry - #35184 Expose executor skills through skill tools @copyberry - #35194 Preserve output from hooks that exit before reading stdin @copyberry - #35196 Make the Apps recovery exposure test deterministic @copyberry - #35198 Enable resource reads for explicit executor skills @copyberry - #35204 Refresh MCP runtimes across thread startup @copyberry - #35205 Use current MCP authority for elicitation reviews @copyberry - #35213 Refresh managed MCP requirements for active threads @copyberry - #35216 Refresh MCP config independently across threads @copyberry - #35220 Support paginated thread forks @copyberry - #35221 Avoid persisting non-local threads for hook transcripts @copyberry - #35238 Support the ent26 enterprise plan @copyberry - #35239 Route MCP auth discovery through runtime HTTP clients @copyberry - #35251 Support ephemeral forks of paginated threads @copyberry - #35254 Expose workspace plugin publish capability @copyberry - #35261 Propagate remote plugin IDs to skill metadata @copyberry - #35262 Track remote plugin IDs in skill invocation analytics @copyberry - #35264 Sign bundled macOS helper binaries @copyberry - #35266 Allow disabling the in-process code-mode host fallback @copyberry - #35267 Harden network approval cancellation and concurrency @copyberry - #35271 Include code-mode tool names in Responses Lite metadata @copyberry - #35275 Trace remote exec-server connection setup @copyberry - #35280 Skip plugin MCP filtering when no allowlists are configured @copyberry - #35359 Handle exec-server network policy requests in the client @copyberry - #35363 Include item start times in completion events @copyberry - #35364 Bound Code Mode metadata compatibility headers @copyberry - #35365 Keep unified mention results fresh @copyberry - #35375 Make the keymap action menu responsive @copyberry - #35408 Ignore generated system skills in the skills watcher @copyberry - #35414 Raise the MCP server recursion limit @copyberry - #35523 Shut down the in-process outbound router explicitly @copyberry - #35524 Preserve terminal turn errors in replayed history @copyberry - #35525 Skip inactive TUI threads without pending user interaction @copyberry - #35530 Track model and personality in world state @copyberry - #35537 Add managed policy for in-app updates @copyberry

Codex

Codex Platform

View all 2 →

Codex Platform

Major

GPT-5.4 and GPT-5.4 mini retire from Codex on August 31

Jul 31, 2026Update soonOfficial source ↗

# GPT-5.4 and GPT-5.4 mini retire from Codex on August 31 On August 31, 2026, GPT-5.4 and GPT-5.4 mini will no longer be available in Codex for users signed in with ChatGPT. GPT-5.4 and GPT-5.4 mini will remain available on the OpenAI API and Codex sessions authenticated with an API key. Switch to their recommended replacements: - Replace `gpt-5.4` with `gpt-5.6-terra` (GPT-5.6 Terra). - Replace `gpt-5.4-mini` with `gpt-5.6-luna` (GPT-5.6 Luna). Before the cutoff, update workspace defaults, saved model settings, managed configurations, custom agents, and scheduled tasks that use either model. See [Codex models](/codex/models#deprecated-codex-models) and [workspace model availability](/codex/enterprise/workspace-model-availability) for details.

Codex Platform

Feature

Sign in with ChatGPT (beta)

Jul 29, 2026OptionalOfficial source ↗

# Sign in with ChatGPT (beta) Sign in with ChatGPT is beginning to roll out in beta across select plugins and partner sites, starting with Airtable, GitLab, HubSpot, Notion, Supabase, and Vercel. When you connect a supported plugin from the ChatGPT plugin directory, you can use Sign in with ChatGPT to create or link an account with that service in fewer steps. On participating partner sites, you can also select **Sign in with ChatGPT** to create or access your account. This makes it easier to start working with supported tools in ChatGPT and Codex. When you sign in, the partner receives only your name, email address, and profile picture, if available. You must still review and approve each plugin's requested access as a separate step.

Claude

Claude Code

View all 20 →

Claude Code

Major

Claude Code 2.1.223

Aug 6, 20262.1.223Update soonOfficial source ↗

- Added owner wildcard entries (`"owner/*"`) to the `strictKnownMarketplaces` and `blockedMarketplaces` managed settings for allowing or blocking all marketplace repos under a GitHub org - Added a warning when workflow agents, forked skills, slash commands, or resumed background agents' requested subagent model is restricted and the parent model runs instead - Added a `/teleport` hint in cloud sessions showing how to continue locally with `claude --teleport <session id>` - Fixed a Bash permission bypass where a crafted command could hide parts of itself from permission checks - Fixed permission prompts so commands padded with tabs or invisible Unicode can no longer hide part of the command from the approval dialog - Fixed workflow scripts being able to use dynamic `import()` to run code outside the workflow sandbox - Fixed a permission gap where an agent definition's `bypassPermissions` mode ignored the org bypass-permissions disable policy - Fixed resuming a session after a mid-session `/cd` coming back empty - Fixed gateway model discovery hiding Claude models registered under provider-prefixed IDs such as `vertex_ai/claude-*` or `bedrock/anthropic.claude-*` - Fixed `modelOverrides` keys that aren't Anthropic model IDs being treated as the session's canonical model ID; unknown keys are now ignored as documented - Fixed managed settings: server-delivered settings no longer disable the env block of a machine-local `managed-settings.json` or MDM profile; admin env now merges per key - Fixed sandboxed commands failing to start on Linux when `sandbox.filesystem.denyWrite` covers the working directory - Fixed forked background agents getting stuck "already resuming" for the rest of the session when rebuilding the fork's parent prompt failed during resume - Fixed a resumed session failing every turn, or leaving the interactive app on an unresponsive error screen, when its history held a malformed diagnostics attachment - Fixed a rare hang when parsing unusual `git push` output - Changed `CLAUDE_CODE_DISABLE_1M_CONTEXT` to hold every Claude model with a native 1M window to 200K via auto-compaction, not just a fixed list; a startup warning now appears when auto-compaction isn't holding the session to 200K - Changed auto-compact to keep sessions on unrecognized model IDs within the assumed context window instead of letting them grow past it; set `CLAUDE_CODE_DISABLE_UNKNOWN_MODEL_WINDOW_ENFORCEMENT=1` to restore the previous behavior - Changed `/review` to be an alias of `/code-review`, which reviews the current diff or a PR (`/code-review <level> <pr#>`); use `/code-review ultra` for a deep cloud review - Changed `/code-review` with no effort level to reuse the level you typed last; type a level like `/code-review high` to change it

Claude Code

Major

Claude Code 2.1.222

Aug 5, 20262.1.222RecommendedOfficial source ↗

- Fixed worktree-isolated sessions and their subagents being able to run destructive git commands against the main checkout; isolation now applies to file edits and Bash in every session type - Fixed PreToolUse auto-allow hooks bypassing tool restrictions in background agent tasks (summaries, compaction, renames) - Fixed `/usage-credits` on Team and Enterprise showing "you've already sent a usage credit request" for members whose earlier request was dismissed, blocking them from sending a new one - Fixed the startup connectivity check hanging and then failing behind an HTTPS proxy; it now uses the same proxy-aware transport as API requests and times out with a clear message - Fixed "Connection closed mid-response" errors being reported on responses that had actually completed - Fixed `/usage` overattributing usage to MCP servers: a server's share now reflects only the requests that actually consumed its tool results, instead of every turn after any call to it - Fixed sessions not linking to pull requests created after the branch was pushed, including through the GitHub REST API - Fixed org-restricted `model: opus`-style subagent and teammate family aliases dropping to the parent model instead of stepping down to the newest org-allowed model in the family - Fixed stream idle timeout firing on custom `ANTHROPIC_BASE_URL` gateways despite server keep-alive pings arriving on the wire - Fixed claude.ai connectors being falsely marked as needing authorization when the session token is invalid — they now show a `/login` hint instead - Fixed tool errors not being displayed for tools no longer available locally, for example after an MCP server is removed - Fixed `SendMessage` rejecting a long summary — it now truncates instead, so sends no longer fail on a character limit - Fixed the spinner's effort label in a subagent's transcript view showing the session's effort level instead of the subagent's own `effort:` setting - Fixed rare crashes when a file watcher hit a filesystem error or during file-watcher teardown - Fixed screen readers re-reading the whole input line on every backspace in `--ax-screen-reader` mode — end-of-line deletions now echo just the deleted characters - Fixed host model-selection keys not taking precedence over a stale on-disk `managed-settings.json` when `CLAUDE_CODE_PROVIDER_MANAGED_BY_HOST` is set - Improved auto mode safety: messages sent to other agent sessions via `SendMessage` are now evaluated by the permission classifier before dispatch - Improved the refusal when Claude tries to invoke a skill with `disable-model-invocation`: Claude is now told to ask you to run the skill instead of replicating its workflow - Improved the `/diff` view, the Remote Control workspace diff, and file-edit diffs in Claude Code on the web sessions to use raw git blob content, ignoring workspace-configured diff drivers and textconv - Changed Remote Control auto-start so repo-local settings (`.claude/settings.json` or `.claude/settings.local.json`) can no longer turn it on (they can still turn it off); enable it at user scope via `/config` - Removed ultraplan feature

Claude

Claude Apps

View all 4 →

Claude Apps

Feature

Claude Opus 5 launch

Jul 24, 2026RecommendedOfficial source ↗

Claude Opus 5 launch We launched Claude Opus 5, a thoughtful and proactive model that comes close to the frontier intelligence of Claude Fable 5 at half the price. For more information, see our blog post: Claude Opus 5.

Claude Apps

Feature

HIPAA configuration for your Claude organizations is now self-serve

Jul 14, 2026RecommendedOfficial source ↗

HIPAA configuration for your Claude organizations is now self-serve You can now manage HIPAA readiness for your Claude organizations yourself. This applies to both Claude Enterprise and the Claude Platform (API). In each product, an eligible admin can review the Business Associate Agreement (BAA), download the implementation guide, and enable the HIPAA configuration in a single flow. For more information, refer to HIPAA-ready Enterprise plans and HIPAA readiness for Claude API.

Claude

Claude Platform

View all 9 →

Claude Platform

Major

Inference hooks are now in beta for Claude Enterprise organizations. Point Claude at your … 等 2 项

Aug 5, 2026Update soonOfficial source ↗

* **Inference hooks** are now in beta for Claude Enterprise organizations. Point Claude at your organization's AI security server, and each governed prompt across claude.ai, Cowork, and Claude Code is held for the server's allow or deny verdict before inference proceeds. Requests are signed, failure handling is configurable, and every denial is recorded in the compliance [Activity Feed](/docs/en/manage-claude/compliance-activity-feed). See [Inference hooks](/docs/en/manage-claude/inference-hooks). - We've retired the Claude Opus 4.1 model (`claude-opus-4-1-20250805`). All requests to this model will now return an error. We recommend upgrading to [Claude Opus 5](/docs/en/about-claude/models/overview#latest-models-comparison). Researchers can request ongoing access through the [External Researcher Access Program](https://support.claude.com/en/articles/9125743-what-is-the-external-researcher-access-program).

Claude Platform

Feature

Dreams (research preview) now supports Claude Opus 5. See Supported models.

Aug 1, 2026RecommendedOfficial source ↗

* [Dreams](/docs/en/managed-agents/dreams) (research preview) now supports Claude Opus 5. See [Supported models](/docs/en/managed-agents/dreams#limits).