Антропік Клод
Швидке розширення штучного інтелекту Google: Gemini 4 Neon і прагнення до агентського домінування
Повідомлені твердження
It wasn't until November (and Claude Fable 5.2) that I felt familiar enough with how it all works to confidently ship it to production! [повідомлене твердження] [D457]
or /cd; Bash may ask first - Changed sandboxed commands on macOS, Linux and WSL to no longer write to some sensitive home folders by default; some can be allowed with sandbox.filesystem.allowWrite in user or managed settings - Changed the Bash sandbox to set aside unexpected entries in a repository's .git folder after a command - Changed the Bash sandbox on Linux and WSL to create some missing --add-dir folders before each sandboxed command, so commands that create them can fail, skip, or land inside them - Changed commands that a permission hook or SDK host rewrites while approving to stay in the sandbox if the original was checked as sandboxed; the approver can add dangerouslyDisableSandbox: true - Changed claude mcp serve to ignore sandbox.excludedCommands - Changed /rewind on macOS, Linux and WSL to leave some files untouched and report them as skipped when the sandbox is on - Changed the Bash sandbox to ask before some lines that set environment variables when auto-allow is off; unattended runs can refuse them - Changed Claude's file tools to refuse writes to Claude Code's sign-in files, such as .credentials.json - Changed the Bash permission check to ask before more commands reach Claude Code's sign-in files - Changed some file tools to refuse some device and process files - Changed some cp and mv commands to ask first - Changed some tool calls in bypass permissions mode to ask for approval, or be denied where no one can be asked - Changed the Bash permission check to a [повідомлене твердження] [D1077]
or /cd; Bash may ask first - Changed sandboxed commands on macOS, Linux and WSL to no longer write to some sensitive home folders by default; some can be allowed with sandbox.filesystem.allowWrite in user or managed settings - Changed the Bash sandbox to set aside unexpected entries in a repository's .git folder after a command - Changed the Bash sandbox on Linux and WSL to create some missing --add-dir folders before each sandboxed command, so commands that create them can fail, skip, or land inside them - Changed commands that a permission hook or SDK host rewrites while approving to stay in the sandbox if the original was checked as sandboxed; the approver can add dangerouslyDisableSandbox: true - Changed claude mcp serve to ignore sandbox.excludedCommands - Changed /rewind on macOS, Linux and WSL to leave some files untouched and report them as skipped when the sandbox is on - Changed the Bash sandbox to ask before some lines that set environment variables when auto-allow is off; unattended runs can refuse them - Changed Claude's file tools to refuse writes to Claude Code's sign-in files, such as .credentials.json - Changed the Bash permission check to ask before more commands reach Claude Code's sign-in files - Changed some file tools to refuse some device and process files - Changed some cp and mv commands to ask first - Changed some tool calls in bypass permissions mode to ask for approval, or be denied where no one can be asked - Changed the Bash permission check to a [повідомлене твердження] [D1077]
question in /ide, where Ctrl+C twice quit Claude Code - Fixed first-run terminal setup saying nothing about what it changed or could not set up - Fixed /status showing installation warnings twice - Fixed the footer's PR badge showing the previous branch's pull request after switching branches - Fixed the "Pushed to" summary naming the fetched branch when a command fetched or pulled first, and "Merged PR" or "Closed PR" showing when gh refused the action and the exit code was hidden - Fixed URLs that wrap over several lines not being clickable in terminals with link support that Claude Code doesn't recognize, and labeled links in the Rex terminal - Fixed the Ctrl+O transcript hinting Ctrl+B to background a command in fullscreen, where it pages up, and showing no sign that Claude is still working - Fixed the Ctrl+R search, the model picker, plan approval, question previews, background shell output and /plugin install choices being cut off in short fullscreen windows - Fixed the /context grid, API errors such as "Credit balance too low", "Waiting for API response" and retry notices, "esc to interrupt" and "Interrupted" being cut off or jumbled in narrow terminals - Fixed wide tables breaking apart in plans and subagent prompts, cutting short words in half beside a long cell, and showing stray words and blank rows when drawn as label and value lists - Fixed code keywords, added diff lines, the slow-reply spinner and "thinking" being hard to read in some themes and terminals - Fix [твердження джерела] [D1077]
twice while a slow backup copy was still running - Fixed /rewind saying "No code changes" when a shell command changed files it can't restore; a turn whose changes weren't tracked now reads "Code changes unknown" - Fixed /rewind: Esc now cancels its summary, its list shows more than one past prompt in fullscreen, and a long prompt no longer hides its warning about losing later file changes - Fixed Ctrl+C or Ctrl+D twice in /skills quitting Claude Code and dropping changes; it now saves and closes the panel - Fixed the usage limit menu erasing your draft and opening while you were typing - Fixed pasted text and images being dropped when ↑ recalls a draft cleared with Esc Esc - Fixed an edited typed answer to Claude's question sending the old text when you moved on without pressing Enter - Fixed going back to a question you answered from its list putting the cursor on the first option, where Enter replaced your answer - Fixed /feedback and /bug sending the report when you typed a space on the confirmation screen, and losing the typed report after Ctrl+O or Ctrl+Z - Fixed a prompt typed after --add-dir, --allowedTools or another list option being dropped without a message - Fixed /plan [повідомлене твердження] [D1077]
Added an orchestrator metric, claude_code_self_hosted_orchestrator_sessions_given_up_total, that counts by cause how often Anthropic gave up starting a runner for a session - Fixed Bash commands in zsh stopping with a "not found" error at a word that starts with =, such as echo ==== or [ a == b ]; =cmd no longer expands to a command's path - Fixed automatic compaction reusing the session's earlier system prompt instead of rebuilding it the way /compact does - Fixed markdown rendering freezing on some very long or malformed input - Fixed rare screen freezes - Fixed Esc often being lost at a permission prompt or question when you typed straight after it; in terminals that send Alt as Esc and a key, an unused Alt or Option key now declines there - Fixed a paste or a terminal's reply being typed into the prompt when a slow connection cut it in two - Fixed saving a scheduled task erasing the project's other saved tasks when the scheduled tasks file can't be read, on third-party providers or with telemetry or feature flags turned off - Fixed a scheduled task with a typo in the project's scheduled tasks file being deleted the next time that file was saved - Fixed a case where a resumed cloud session could lose its compaction summary or recent messages when a message was saved late - Fixed the token counts on the spinner and for subagents counting a retried reply twice, missing replies that were not streamed, and standing still while Claude thinks - Fixed the error shown when usage c [повідомлене твердження] [D1075]
and most /config value lists to be unnumbered outside screen reader mode; use the arrow keys and Enter - Changed Bash commands whose variables expand to very long text to be denied, with a message that says to use shorter values or a file - Changed the PowerShell tool to ask before one more command, even under a rule naming it; if nothing answers, it's refused unless a whole-tool rule, auto mode or bypass permissions mode allows it - Changed a subagent resumed by a sibling's message to follow the session's permission settings; it may now ask, or be refused where nobody can answer - Changed --worktree with a relative --add-dir beside the launch folder to give the Bash sandbox that folder, not one beside the worktree - Changed some --worktree starts with a relative --add-dir above the launch folder, such as .., to give the Bash sandbox, and a --bg session, only the worktree - Changed Claude in Chrome to refuse uploading some credential files; attach those to the page yourself - Changed artifact approval in organizations that require it for each new artifact: an artifact that only a plugin created or allowed can be refused - Changed sessions with nobody to ask to refuse a new artifact at once, with no permission prompt, in organizations that require approval of each one - Changed the Artifact tool to refuse calls once settings turn Artifacts off, in sessions already running too - Changed the Artifact publish prompt to stop offering "don't ask again" in subagents and after a back [повідомлене твердження] [D1076]
help hiding how to exit in a short window - Fixed Ctrl+C twice not closing /chrome from its browser list - Fixed a number and Enter typed too early at an agent's plan approval in screen reader mode causing the next number to be dropped too - Fixed Claude trying to read design system token files that do not exist when starting a deck or design - Fixed exit not saying that a worktree was kept - Fixed /hooks help naming input fields that do not exist - Fixed question dialogs naming ctrl+g after the external editor key was rebound - Fixed two sign-in error messages that pointed to the Claude subscription option on computers where forceLoginMethod is set to console - Fixed /config saying the dialog was dismissed instead of "Set theme to …" after changing the theme - Bug fixes and reliability improvements - Improved the speed of a session's first Bash command: shell start-up files may now run before it's approved, even if it's refused; set CLAUDE_CODE_DISABLE_SHELL_HEAD_START=1 to turn this off - Improved commits on newer models and in PowerShell: Claude is now told not to commit files that likely contain secrets - Improved sandboxed shell commands: Claude now sees when a command exempt from the sandbox stayed in it because its permission check relied on the sandbox - Improved the refusal for shell commands with no rm that the dangerous rm check can't parse, so Claude can resend harmless ones in plain syntax instead of giving up - Improved resuming very long sessions: a transcript [повідомлене твердження] [D1076]
Failure-loop intervention watches for an agent that keeps hitting the same failure. [повідомлене твердження] [D650]
Швидке розширення штучного інтелекту Google: Gemini 4 Neon і прагнення до агентського домінування [опубліковане повідомлення] [N12]
Хронологія
- AI Gateway now supports Private Inference [D714]
- Post-Training NVIDIA Nemotron to Increase Accuracy in Your Scientific Domain [D662]
- Hy4 Flash now available on AI Gateway for 50% off [D713]
- TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient [D650]
- v2.1.326 [D1077]
- AWS announces agentic account creation for agentic coding harnesses [D1419]
- Running the full server-side Datasette Python app in the browser with service workers and Pyodide [D457]
- v2.1.327 [D1076]
- Google представила Gemini 4 Neon: нова ШІ-модель посіла перше місце в рейтингу ШІ для програмування [D1850][D1563][D697]
- $3.26 million in SpaceX AI data center revenue will be pumped back into the City of Memphis [D546]
- Hugging Face now routes model requests through OpenRouter for the best price or speed [D508]
- Anthropic, OpenAI, and Google models readily marked medics, journalists, and surrendering soldiers as kamikaze drone targets [D505]
- How Stretto built an AI voice agent with Amazon Nova 2 Sonic [D1282]
- How Social Department built culture-aware video discovery [D1271]
- v2.1.328 [D1075]
- automodel: A Skill for Agentic Model Discovery [D986]
Першоджерела
- [D986] · arXiv cs.AIautomodel: A Skill for Agentic Model Discovery
- [D1075] · Claude Code Releasesv2.1.328
- [D500] · The DecoderGoogle ships Gemini 4 Neon, its fourth new Gemini model in a month
- [D1271] · AWS Machine LearningHow Social Department built culture-aware video discovery
- [D546] · Tom's Hardware$3.26 million in SpaceX AI data center revenue will be pumped back into the City of Memphis
- [D457] · Simon WillisonRunning the full server-side Datasette Python app in the browser with service workers and Pyodide
- [D84] · PsyPost PsychologyScientists who use AI propose less distinctive ideas for federal research grants
- [D650] · Microsoft ResearchTokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient
- [D713] · Vercel ChangelogHy4 Flash now available on AI Gateway for 50% off
- [D662] · NVIDIA Developer BlogPost-Training NVIDIA Nemotron to Increase Accuracy in Your Scientific Domain
- [D1077] · Claude Code Releasesv2.1.326
- [D1076] · Claude Code Releasesv2.1.327
- [D1282] · AWS Machine LearningHow Stretto built an AI voice agent with Amazon Nova 2 Sonic
- [D505] · The DecoderAnthropic, OpenAI, and Google models readily marked medics, journalists, and surrendering soldiers as kamikaze drone targets
- [D508] · The DecoderHugging Face now routes model requests through OpenRouter for the best price or speed
- [D1850] · Dzerkalo TyzhdnyaGoogle представила Gemini 4 Neon: нова ШІ-модель посіла перше місце в рейтингу ШІ для програмування
- [D1563] · Simon Willisonllm-gemini 0.38
- [D697] · Vercel ChangelogGemini 4 Neon now available on AI Gateway
- [D1419] · AWS What's NewAWS announces agentic account creation for agentic coding harnesses
- [D714] · Vercel ChangelogAI Gateway now supports Private Inference
- [N12] · Google ships Gemini 4 Neon, its fourth new Gemini model in a monthGoogle ships Gemini 4 Neon, its fourth new Gemini model in a month