Codex 업데이트 기록

Radar Ultra

릴리스 217개

0.96.0

CLIGitHub 원문

AI 요약

  • Thread listings now check the state database first, including archived threads, and use filesystem traversal only when needed. This improves listing correctness and resilience.
  • On non-Windows platforms, unified_exec is now enabled.
  • Websocket clients can receive rate-limit signals through a new event, with ETag and reasoning-metadata handling matching websocket behavior.
  • Clients can start thread compaction through a new API call and track its completion separately; it runs asynchronously.
  • The TUI’s Esc key now exits notes mode instead of interrupting the session when notes are open in the request_user_input overlay.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.95.0

CLIGitHub 원문

AI 요약

  • Git commands that can change or delete data are now less likely to bypass approval checks.
  • You can launch Codex Desktop on macOS with codex app <path>; Codex downloads the DMG automatically if needed.
  • Shell-related tools can run in parallel. Shell executions also receive CODEX_THREAD_ID, which scripts and skills can use to identify the active session.
  • You can load personal skills from ~/.agents/skills (with compatibility for ~/.codex/skills), and app-server APIs and events can list and download public remote skills.
  • /plan accepts prompt arguments and pasted images. The release also fixes interrupted codex exec runs that could hang in websocket or streaming flows.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

26.204

App공식 원문

AI 요약

  • You can now open files and folders in Zed or TextMate.
  • The review panel can now preview PDFs.
  • Performance has been improved.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

26.203

App공식 원문

AI 요약

  • You can now rename a thread by double-clicking it in the thread list.
  • Sync is now called Handoff, and its interface shows clearer source and destination stats.
  • The release also includes unspecified performance improvements and bug fixes.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.94.0

CLIGitHub 원문

AI 요약

  • Plan mode is now on by default, with updated guidance in its prompt.
  • Skills can now be loaded from .agents/skills; instructions for relative paths are clearer, and nested folders can be marked as skills.
  • Personality settings are now stable: the default is friendly, the setting is named personality, and existing settings migrate forward.
  • Console output now includes runtime metrics to help with diagnostics.
  • Unarchiving a thread refreshes its timestamp and sidebar ordering; conversation rules are capped and duplicate prefix rules are removed; override turn context no longer adds extra items.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.93.0

CLIGitHub 원문

AI 요약

  • Smart approvals are enabled by default, and MCP tool calls now show explicit approval prompts.
  • Plan mode now streams proposed plans in a dedicated TUI view; a feature-gated /plan shortcut lets you switch modes quickly.
  • Use /apps to browse connectors in the TUI and $ to insert apps into prompts.
  • The app server can accept ChatGPT authentication tokens from a host app and request refreshed tokens when needed.
  • Several fixes improve everyday use: file search follows session folder changes and handles multiple roots; input history restores image attachments and rich text; and connections fall back to HTTP if WebSocket proxies fail.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.92.0

CLIGitHub 원문

AI 요약

  • Collaboration flows use less CPU, and new safeguards limit how deeply agents can collaborate. They also add an explorer role and improve collaboration event handling.
  • The web search tool now shows activity while it works and handles all action types. Cached web search is now the default client behavior.
  • You can filter the app’s thread list to browse large collections more easily, and restore archived threads to active sessions with a new RPC.
  • MCP servers can define OAuth scopes in config.toml, so you need to pass --scopes on login less often.
  • Fixed a TUI freeze under heavy streaming, and corrected codex resume --last --json prompt parsing.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.91.0

CLIGitHub 원문

AI 요약

  • The maximum number of sub-agents is now 6, tightening limits on how many can be used at once.
  • This change is intended to tighten resource usage and guardrails for agent fan-out.
  • No other changes are described in these notes.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.90.0

CLIGitHub 원문

AI 요약

  • A network sandbox proxy now helps control outbound network access through policy enforcement.
  • Collaboration mode is available as a TUI beta, with a clearer plan-to-execute handoff and simpler Coding or Plan mode selection.
  • Connectors support has begun through the app server and MCP integration, with related configuration and documentation updates.
  • Resume is more reliable: Codex handles out-of-order events and prompts for the working directory when it differs.
  • Several usability fixes make errors and defaults more dependable: configuration errors are clearer, filtered model presets are respected, and --yolo no longer fails outside a Git repository.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.89.0

CLIGitHub 원문

AI 요약

  • You can now use /permissions for a shorter approval workflow; /approvals remains available for compatibility.
  • A new /skill screen lets you turn individual skills on or off.
  • Slash-command suggestions now favor exact and prefix matches over fuzzy matches.
  • The app server can read individual threads and filter archived threads from thread lists. Its clients also support layered configuration files, and config/read can calculate effective settings for a specified working directory.
  • Tilde paths such as ~//... can no longer expand outside your home directory.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.88.0

CLIGitHub 원문

AI 요약

  • Headless environments now have device-code authentication as a standalone fallback.
  • Configuration is loaded only from trusted folders, and symlinked config.toml files are now resolved correctly.
  • Interrupted turns no longer repeat.
  • Fixed Azure endpoint input errors, a memory leak in the core runtime, and an image-paste regression in the WSL interface.
  • Updated MCP documentation links and collaboration-mode developer instructions, and corrected a naming typo.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.87.0

CLIGitHub 원문

AI 요약

  • Piped commands that don’t use a terminal no longer hang while waiting for input.
  • User shell commands now use your shell settings and aliases.
  • The TUI now shows approval requests from spawned or unsubscribed threads.
  • Token estimates during compaction are more accurate, improving budget estimates in long sessions.
  • User-message annotations can now survive history rebuilds, and MCP tool results include a thread ID and defined output schema for compatibility.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.86.0

CLIGitHub 원문

AI 요약

  • Skill metadata can now be set in SKILL.toml, including names, descriptions, icons, brand color, and a default prompt, and is surfaced in the app server and TUI.
  • Clients can explicitly disable web search and send an eligibility header, supporting alignment with server-side rollout controls.
  • Accepting an MCP elicitation now sends an empty JSON payload instead of null, for servers that expect content.
  • Several interaction and process issues were fixed: prompt placeholders are no longer italic, empty paste events no longer read clipboard images, and Unified exec cleans up background processes to prevent late End events.
  • Internal changes refresh the orchestrator prompt to improve routing behavior and reduce noisy needs_follow_up error logs.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.85.0

CLIGitHub 원문

AI 요약

  • App-server v2 now sends collaboration tool calls as turn-stream events, letting clients show agent coordination as it happens.
  • You can assign a role preset when spawning an agent, and optionally interrupt a running agent before sending it a message.
  • codex resume --last now resumes sessions from the current working directory; use --all to override that limit.
  • Model upgrade suggestions can include richer guidance from migration notes, when the client displays the new metadata.
  • Linux sandboxing now uses Landlock-only restrictions if user namespaces are unavailable, and sets no_new_privs before applying sandbox rules.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.84.0

CLIGitHub 원문

AI 요약

  • Text elements can now include additional metadata, allowing clients to render them in richer ways and making the schema easier to evolve.
  • Rust release builds have a longer job timeout to reduce flaky pipeline failures, particularly on Windows.
  • Rust release builds have a longer job timeout to reduce flaky pipeline failures, particularly on Windows.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.81.0

CLIGitHub 원문

AI 요약

  • The default API model is now gpt-5.2-codex.
  • Headless runs now switch to device-code sign-in automatically, so you can sign in without a browser.
  • On Linux, the sandbox can mount paths read-only to better protect files from changes.
  • The codex tool in codex mcp-server now returns a threadId that can be used with codex-reply.
  • Several issues were fixed: Zellij scrollback handling, unsafe-command prompts on Windows with a read-only sandbox, reporting configuration and rules parsing errors, a macOS proxy-discovery crash, and errors for invalid image uploads.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.80.0

CLIGitHub 원문

AI 요약

  • CLI subprocesses again inherit important environment variables, fixing runtime problems that could cause 10× or greater performance regressions, especially for GPU-related subprocesses.
  • You can now branch a session into a separate thread using new conversation-fork endpoints.
  • Codex now offers elevated-sandbox onboarding, including prompts for upgrading or continuing in degraded mode and an /elevate-sandbox command.
  • In the TUI and TUI2, /review <instructions> now starts the review flow instead of being sent as ordinary text.
  • Other updates add explicit skill invocation through the v2 API, expose requirements for clients, and add observability counters.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.79.0

CLIGitHub 원문

AI 요약

  • Sessions can now spawn or message other conversations, making it possible to coordinate work across multiple conversations.
  • IDE clients can undo the last several turns in a conversation without rewriting its history.
  • Web Search can use cached results only, as an alternative to making live requests.
  • A production policy bypass was closed: CODEX_MANAGED_CONFIG_PATH can no longer override managed configuration in production.
  • Several usability and reliability fixes improve transcript resizing and copy/paste, tolerate extra whitespace in apply_patch markers, and reduce TUI glitches.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.78.0

CLIGitHub 원문

AI 요약

  • You can press Ctrl+G to edit your current prompt in your configured external editor; edits sync back into Codex.
  • Codex can combine project-level .codex/config.toml settings with system configuration. Project-root detection can use configurable markers, and enterprise-managed configuration requirements are supported on macOS through an MDM-provided TOML payload.
  • Multiple tool calls returned by a model in one turn are now handled in the expected format, fixing failures in those cases.
  • The tui2 interface has improved transcript selection, copying, and scrolling, along with fixes for rendering and input edge cases.
  • Windows PowerShell sessions now start in UTF-8 mode to reduce encoding-related prompt and output issues.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

0.77.0

CLIGitHub 원문

AI 요약

  • The /undo command is fixed so it no longer disrupts Git staging or ghost commits.
  • Mouse-wheel and trackpad scrolling are more consistent across terminals, with new settings to adjust scrolling.
  • Requirements can now specify which sandbox modes are allowed, helping constrain the modes Codex may use.
  • OAuth login for streamable HTTP MCP servers no longer requires the rmcp_client feature flag.
  • File search now uses centralized filename handling to improve the display and consistency of fuzzy search results.

공식 노트를 바탕으로 AI가 정리했습니다. 원문이 우선합니다.

Radar Ultra
OpenAI와 관련 없는 비공식 도구입니다. 공개 게시물을 정리한 것이며 원본 게시물이 기준입니다.