Copilot CLI

持续追踪核心框架、最新模型规格与重要系统接口的更新迭代轨迹。

1.0.79-7

1.0.79-7

Added Agent Plugins spec plugins can now ship extensions under a com.github.copilot/extensions/ directory Add support for the kimi-k3 model Combine --plan with --mode autopilot to plan first and then implement without waiting for approval Improved Ask user multi-select prompts include an Other option for free-text answers Improve teleported subagent /tasks navigation with nested tree browsing, current/all and finished-task filters, and a live timeline you can steer Fixed The /app command now opens the current session in the GitHub Copilot desktop app instead of landing on Home with the wrong f...

1.0.79-6

1.0.79-6

Fixed A rare internal delay no longer prints a diagnostic warning on top of the interactive UI A failed session-history load no longer leaves the timeline permanently empty: the failure was silently discarded, so the transcript stayed blank for the rest of the session with nothing logged. It is now retried, and reported in the transcript and the log if it still fails Resuming a long session no longer collapses the timeline's scroll range while history renders in the background: entries that had not finished rendering were published as if they did not exist, so the scrollbar and scroll position...

1.0.79-3

1.0.79-3

Improved Use /worktree new to start a new session in a new worktree

1.0.79-5

1.0.79-5

Added Manage multiple concurrent sessions from the Sessions tab and sidebar Improved Prompt pinning is off by default; set pinnedPrompts to true to enable it. Fixed Sandboxed wrapper builds (make and friends) get the dev tool caches their recipes need, based on the build manifests in the working directory

1.0.79-2

1.0.79-2

Improved Pin the current prompt one row higher, in the row the tab bar already reserves, so it keeps the shape of the prompt it copies while costing the timeline one row less Leave the pinned prompt off by default on terminals under 30 rows, where it would crowd the output; set pinnedPrompts explicitly to override at any size Fixed A sandbox that cannot start an MCP server now fails in seconds instead of stalling the session, and sandbox startup failures for both MCP and language servers now say the sandbox was at fault and how to fix or opt out of it Login links are clickable during web and d...

1.0.79-1

1.0.79-1

Improved BREAKING: the sandbox setting allowDevToolCaches is renamed allowDevToolAccess , since it grants dev-tool config and registries too, not just caches. The old key is no longer read and is ignored silently, so an existing false opt-out reverts to the default (on). Rename it in settings.json and in any managed/MDM policy. Fixed Compute /context attribution against the Auto-resolved model so token totals are accurate for Free/Student users Disabling an extension no longer breaks elicitation, canvases, or tool permission prompts for other extensions A prompt stashed with ctrl+s now stays w...

1.0.78-3

1.0.78-3

Added Add the experimental /new-worktree command to create a new worktree and start a new conversation in it Improved Interactive shell shortcut now launches on Enter and shows an inline hint when "$" is armed Fixed Copilot login now defaults to the browser flow for local desktop subprocesses without a TTY, including IDE integrations, while remote and headless environments continue using device code

1.0.78

1.0.78

2026-08-03 Timeline headers show how long each tool call took, right-aligned and ticking live while it runs (for calls of at least 5 seconds). On by default — disable with /settings showToolDurations . First-party plugins automatically update to the latest version at session start Add the experimental /new-worktree command to create a new worktree and start a new conversation in it Copilot login now defaults to the browser flow for local desktop subprocesses without a TTY, including IDE integrations, while remote and headless environments continue using device code Interactive shell shortcut n...

1.0.78-2

1.0.78-2

Improved Split-view sidebar: the red close confirmation now reads x again to close (or x again to exit CLI on the last session) instead of x close , so a second press is clearly what closes Fixed Extension slash commands run their handler exactly once per invocation when several extensions are loaded Inline images no longer render with their first row repeated down the whole picture after the timeline scrolls A run whose prompt is piped over stdin now treats its sessionEnd hook the same way -p does: the hook fires once per completed agent turn with reason complete (or error if the turn failed)...

1.0.78-0

1.0.78-0

Added Add /permissions to switch between approval modes. ACP mode supports closing sessions with the closeSession request. Improved New sandbox setting allowDevToolCaches (on by default): grants sandboxed builds access to toolchain caches, registries, and installs so builds work without extra setup. Set false to opt out. Honor explicit GitHub MCP toolset/tool config: keep gh-overlap tools and stop steering to the gh CLI when you opt in Warn on startup about unknown top-level keys in user settings.json (e.g. a misspelled setting) instead of silently ignoring them Shell completion for --model no...

[ 重大突破 ] 1.0.77

1.0.77

2026-07-30 Unconditional autopilot approval now disables sandbox for the current session when bypass is allowed Ctrl+G opens your editor to edit ask_user freeform answers without closing the prompt Add a browser-based (web) OAuth login flow, now the default for copilot login on local interactive terminals (device code remains the default on remote/headless terminals). Use --web-flow / --device-code to force a mode, or pick one in the interactive /login command Support enforcing managed sandbox policy via macOS and Windows native MDM settings Allow reasoning effort to be omitted so the server c...

[ 重大突破 ] 1.0.77-0

1.0.77-0

Added Add a browser-based (web) OAuth login flow, now the default for copilot login on local interactive terminals (device code remains the default on remote/headless terminals). Use --web-flow / --device-code to force a mode, or pick one in the interactive /login command Support enforcing managed sandbox policy via macOS and Windows native MDM settings Improved Allow reasoning effort to be omitted so the server can select the default

1.0.76-4

1.0.76-4

Fixed Sandbox denied paths are enforced for relative and symlinked entries on macOS and Linux (Windows cannot deny per path)

1.0.76-5

1.0.76-5

Added Add enable/disable controls in /plugins for plugins, instructions, agents, LSP servers, and hooks Add support for the grok-4.5 model

1.0.76

1.0.76

2026-07-29 Add enable/disable controls in /plugins for plugins, instructions, agents, LSP servers, and hooks Add support for the grok-4.5 model Sandbox denied paths are enforced for relative and symlinked entries on macOS and Linux (Windows cannot deny per path) Unsent prompt text now stays with the session it was typed for (for the rest of the CLI session) instead of following you to the session you switch to Resuming a session now restores its autopilot or plan mode instead of reverting to interactive, so the autopilot-only task_complete tool stays available and the mode matches the session ...

1.0.76-3

1.0.76-3

Improved When an update is auto-downloaded, the notification suggests /restart and drops the warning color /diff scrolls and syntax-highlights large multi-file diffs faster Split-view sidebar: hover-to-focus is now off by default (opt in with sidebar.hoverFocus ), the active session card is accented by default (opt out with sidebar.accentActiveSession ), and the closed-state open sidebar hint always renders in the neutral hint color web_fetch now follows HTTP redirects instead of failing, asking permission for the redirect target when it is on a different origin and showing where the redirect ...

1.0.76-1

1.0.76-1

Added Voice mode pauses playing media before recording and resumes it afterward, where supported (macOS and Windows) Show the number of active scheduled prompts in the footer Add /limits predict to suggest a session AI-credit limit from similar sessions. Add configurable timed refreshes for custom status-line commands Improved web_fetch routes through the configured sandbox proxy when outbound is allowed, and denies egress when network.allowOutbound is false (a proxy no longer overrides the user's outbound policy); when a proxied fetch fails it warns that curl/wget share the same proxy, and su...

1.0.76-2

1.0.76-2

Added Add a directable queue manager (staff) to reorder, edit, remove, repeat, and immediately send queued messages New Sessions sidebar for managing multiple concurrent sessions: switch between them, spawn new ones, and see their status at a glance. Turn it on with experimental mode ( /experimental on ). Enterprise administrators can enforce a restrictive sandbox floor: managed settings tighten (but never loosen) the user's sandbox policy, and the /sandbox dialog surfaces the org-configured managed values with locked fields and managed filesystem paths so admins can confirm what is enforced. ...

1.0.74-4

1.0.74-4

Added Add support for Open Plugin Spec v1 plugin manifests and mcp.json configuration Improved Subagent timelines identify whether prompts came from the main agent or another subagent Fixed IDE integration reconnects reliably when the CLI reloads MCP servers or changes directory Multi-turn subagent timelines show every prompt and response in the correct order after reopening /tasks

1.0.74

1.0.74

2026-07-23 Typing ? while the /search bar is open enters it as text instead of opening quick help Add support for Open Plugin Spec v1 plugin manifests and mcp.json configuration IDE integration reconnects reliably when the CLI reloads MCP servers or changes directory Multi-turn subagent timelines show every prompt and response in the correct order after reopening /tasks Subagent timelines identify whether prompts came from the main agent or another subagent Show a first-run splash to opt into the default sandbox Adding support for gemini-3.6-flash The /mcp add and /mcp edit wizard now preserve...

1.0.76-0

1.0.76-0

Improved MCP tools load faster from definition-scoped snapshots, with process-wide and per-server cache opt-outs. Autopilot stays selected after task_complete by default; set stayInAutopilot to false to return to interactive mode after each task Fixed Restore the early warning when unreclaimable system and tool context nears the limit, before automatic compaction is blocked Session working directory no longer reverts to the original checkout shortly after /worktree switches into a new worktree

1.0.74-1

1.0.74-1

Added Show a first-run splash to opt into the default sandbox Adding support for gemini-3.6-flash Improved When multiplexing sessions, a session's open dialog no longer leaks into another session; eligible pickers reopen when you switch back The $ interactive shell shortcut now opens a shell even while the agent is working Fully honor the skill disable-model-invocation flag Warn when a participating language server reports a different symbol than the one requested Steering interrupts shell output waits without stopping the running command Increase the Responses request size limit Plan mode now...

1.0.74-0

1.0.74-0

Added Add /model plan (or /model --plan ) to pick a model used while in plan mode; pass a model id, off to clear, or no id to open the picker. Reverts to the session model when you leave plan mode. Improved Resume search matches session titles even when whitespace differs

1.0.73

1.0.73

2026-07-20 Anthropic subagents continue working when additional directories are configured Resolve relative links in custom agent instructions from the agent file location

[ 重大突破 ] 1.0.72

1.0.72

2026-07-20 An agentStop hook that always blocks no longer loops indefinitely: the CLI now ends the turn after 8 consecutive blocks, and agentStop hooks receive a stop_hook_active flag so they can detect a forced continuation and self-limit Add opt-in git and gh authentication inside the OS sandbox Sandbox macOS keychain access now defaults off for tighter isolation; re-enable it in /sandbox if a command needs it Lifecycle and subagent hook commands run in the current session directory after /cd Deleting an MCP server with /mcp delete stops its running background process Toggling /sandbox resta...

1.0.72-0

1.0.72-0

Added Multi-turn subagents are always enabled, so you can send follow-up messages to running agents Enable tool search for Claude Haiku 4.5+ Improved Deliver scheduled prompts as steering messages when the agent is busy Fixed Emoji shortcodes like 🎉 no longer render with a spurious trailing space in printed and PR/issue/gist output

1.0.72-1

1.0.72-1

Added Add --plugin, --mcp, and --skill flags for plugin mutations Add skill removal support to copilot plugins remove --skill Improved Reveal full file paths when expanding compact editing rows Make the plan-approval menu deterministic across models Keep /add-dir directories visible in the agent context across turns Fixed Wrap ask-user and elicitation inputs in the split-pane chat view Modified vim keys (Ctrl+K, uppercase J/K) no longer move the selection in tool-permission prompts and other text-input select menus; only unmodified j/k, the arrow keys, and Ctrl+P/Ctrl+N navigate. /terminal-set...

1.0.71-3

1.0.71-3

Fixed On startup, an invalid settings.json now shows a warning identifying the offending value instead of silently ignoring your settings /terminal-setup no longer skips setup on terminals without real kitty keyboard support

[ 重大突破 ] 1.0.71

1.0.71

2026-07-16 copilot -p --autopilot no longer hangs when a background shell or agent outlives the turn; it now honors the COPILOT_TASK_WAIT_TIMEOUT_SECONDS timeout the same way plain -p does. Reopening the /subagents model picker keeps each agent's reasoning effort and context tier Refresh memory context after 30 minutes in long-lived sessions Keep MCP tool lists up to date when servers change Avoid leaving long-running background git processes after exit Add a configurable maximum for Ctrl+R command history On startup, an invalid settings.json now shows a warning identifying the offending value...

[ 重大突破 ] 1.0.71-1

1.0.71-1

Added Persist GitHub MCP toolset/tool config via settings.json (githubMcpToolsets, githubMcpTools, etc.) Add plugins marketplace subcommands to list, add, and remove plugin marketplaces Persist sidebar sessions across restarts Add plugins marketplace browse and update commands Split /worktree and /move: /worktree now creates a new worktree and leaves your uncommitted changes behind, while the new /move carries them into the new worktree Add local and cloud cost profiles to /chronicle cost-tips Improved The /model picker shows the Auto model description as markdown with a clickable Learn More l...

1.0.71-2

1.0.71-2

Added Add /voice devices to choose and persist the microphone for voice mode Limit which built-in agents are available to tasks and subagents Add canvas support in the CLI for extension-driven interactions Improved Improve /chronicle cost-tips recommendations with richer cost profiles Highlight standalone hex colors inline in Markdown Fixed Enforce the sandbox filesystem policy on LSP file reads and rename edits Allow empty owner and author emails in marketplace metadata Keep all MCP Server Type options visible on short terminals Mark disabled skills in copilot skill list and its JSON output P...

1.0.71-0

1.0.71-0

Added Add a pinned prompts setting in /settings to control prompt pinning Add Repo and Repo (local) scope tabs to the /settings dashboard Improved Use targeted validation commands and lighter install guidance by default Use ctrl+x → x to close a session and ctrl+x → h to hide the split sidebar Fixed Interactive shell commands now fail with a retryable "reconfiguring" message instead of an "unknown shellManager handle" error after the shell manager is disposed, and a detached command's completion notification is no longer lost when the shell context is reconfigured while a read is in progress R...

[ 重大突破 ] 1.0.70

1.0.70

2026-07-09 Add GPT-5.6 model support Show a single Error prefix for mcp and skill command failures Show the real parse error when --agent selects a malformed custom agent web_fetch works through mandatory HTTPS proxies Hide / search on the Gists tab Treat superseded subagent runs as cancellations instead of failures Add paginated session.mcp.resources read/list/listTemplates RPCs for MCP server resources preToolUse hooks that exit with code 2 deny tool calls Create draft skills when Forge finds a clear workflow pattern Hide the GitHub App install nudge in remote terminals Pin plugins to an exa...

[ 重大突破 ] 1.0.70-0

1.0.70-0

Added Pin plugins to an exact commit SHA using the sha field in plugin source configuration Add --sandbox and --no-sandbox flags to turn the OS-level shell sandbox on or off for the current session only, without changing your saved sandbox setting (useful with -p) Add /refine to rewrite a rough, stream-of-consciousness prompt into a clear one Add --repo and --local flags to /settings and /model Add a setting to show or hide timeline timestamps Let a trusted repository pin the model, effort level, and context tier and extend the URL/MCP/skill deny lists via .github/copilot/settings.json Expose ...

[ 重大突破 ] 1.0.69-3

1.0.69-3

Added Let built-in file edits bypass the sandbox when you approve it web_fetch now follows the active sandbox network policy (denying outbound or local targets the policy blocks) and, when the host opts in via sandbox.allowBypass, lets you approve a one-time bypass from the fetch prompt Show exact local assistant usage in Chronicle and session SQL Improved Improve /sandbox add-path suggestions so they open only when you ask for them Improve /chronicle cost-tips with more precise evidence-backed recommendations Hold Alt (Option on macOS) while scrolling to move one line at a time Resume and swi...

[ 重大突破 ] 1.0.69

1.0.69

2026-07-07 Label built-in file edits with a (sandbox policy) badge instead of (sandboxed), since they follow the sandbox policy on a best-effort basis rather than running in the OS-level sandbox Reload installed plugin extensions without restarting a session Add a /plugins dashboard to manage installed plugins Open quick help from an empty prompt without leaving a stray ? Add minimal reasoning effort for gemini-3.5-flash Show a scrollbar in the /model picker list Show disabled servers as disabled in /lsp test Let built-in file edits bypass the sandbox when you approve it web_fetch now follows ...

1.0.69-2

1.0.69-2

Added Show /rubber-duck in pre-auth help and self-documentation Improved Sign in to MCP servers through the CLI OAuth callback flow Reveal the full /user switch picker when the timeline is full so its hint bar is no longer clipped below the terminal Fixed Include files inside new untracked directories in /diff local edits A prompt taller than the screen stays scrollable when its top scrolls off, instead of collapsing to a pinned header Require Copilot login before ACP authenticate returns success Show the full assistant response in prompt mode when --stream is off Disconnect a repo-scoped plug...

[ 重大突破 ] 1.0.69-1

1.0.69-1

Added Add /mcp list to show attached MCP servers and their status, and allow /mcp list and /plugin list to run while the agent is working Allow opening the /mcp manager while the agent is working to enable or disable servers mid-turn; add, edit, delete, and re-auth stay paused until the turn finishes Add auto allow-all mode that auto-approves requests an LLM judge evaluates as acceptable Add a stayInAutopilot setting (default false) that keeps the CLI in autopilot mode after an autopilot task completes Improved Browse and filter models more easily in the CLI model picker Resume sessions faster...

1.0.69-0

1.0.69-0

Added Add file and folder completion to /sandbox path entries Fixed Update a backgrounded session's branch label in the Sessions split view when its working directory changes Skip unnecessary MCP reloads when returning to an already-loaded session Prevent the tgrep indexer from running out of memory on large monorepos, and fall back to ripgrep if the indexer is killed instead of repeatedly restarting it

[ 重大突破 ] 1.0.68

1.0.68

2026-07-01 Add support for the kimi-k2.7-code model The focused field in the /mcp config form is marked with a "❯ " chevron, not color alone Keep IDE tools available during transient IDE disconnects, returning a clear error while disconnected and recovering automatically when the IDE reconnects Tab completion shows slash command aliases inline (e.g. /pr automerge|agentmerge) Stop hooks from erroring and denying every tool when a session's working directory or git worktree has been deleted Keep the footer model status in sync when reasoning effort or context tier changes Avoid duplicate skill a...

1.0.67

1.0.67

2026-06-30 Disabling the sandbox for the rest of the session now takes effect immediately, so shell and search commands stop re-prompting to bypass it mid-turn Subagent sessions keep parent tool restrictions Show warnings and errors when host custom agents fail to load Require session limits to be at least 30 AI credits Add Claude Sonnet 5 as a supported model Allow tool calls to continue when hooks time out Ctrl+Q now enqueues the highlighted slash-command argument completion MCP OAuth against Microsoft Entra servers behind a tenant vanity domain (e.g. Copilot Studio) no longer fails to refre...

[ 重大突破 ] 1.0.66

1.0.66

2026-06-30 Use a non-blinking block cursor during interactive sessions, restoring your terminal's default cursor on exit Add support for Claude Opus 4.8 Fast and deprecate Claude Opus 4.6 Fast MCP add/edit form accepts HTTP-style Key: value headers Keep LSP servers from starting twice during startup Avoid blocking commands that contain Windows-style path fragments Let Copilot read output from and stop detached background shell commands Large output handling now respects custom output directories and a disable setting Prevent PR description generation from crashing on empty assistant responses ...

1.0.66-2

1.0.66-2

Added Allow skills with the same name from different plugins to coexist Let integrations read and write CLI user settings View LSP server logs in /lsp logs and read_agent Prompt to install gh CLI when it is missing in GitHub repositories Add GitHub attachment variants to prompt rendering Improved Perform tokenizer warmup in parallel on a background thread for better startup performance Show submit times next to user prompts in the timeline Improve /share to manage synced session visibility Expand @-style imports in AGENTS.md, CLAUDE.md, and Copilot instruction files Make /pr auto keep working ...

1.0.65

1.0.65

2026-06-24 /cd now persists the working directory so resuming a session returns to it, and discovers custom agents in the new directory Commands with slash-prefixed string arguments (e.g. --body "/azp run") no longer trigger spurious filesystem permission prompts Fullscreen timeline stays anchored when older content is trimmed Resume open canvases automatically after restarting the CLI Add an opt-in status bar item showing CI check status (passing/running/failing) for the current branch Add a copilot skill subcommand (and a /skill alias for /skills ) to list, add, and remove skills from a file...

[ 重大突破 ] 1.0.66-0

1.0.66-0

Added Add toggle to enable or disable MCP servers in the CLI from MCP list view Add experimental response budget controls to CLI settings Let managed settings configure OpenTelemetry export MCP tools on OAuth-authenticated remote servers now recover automatically after a mid-session token expiry, matching the existing OIDC retry behaviour. A 401 during a tool call triggers a non-interactive reconnect, and servers needing interactive re-auth are retried at the start of the next turn. Add persisted dynamicRetrieval setting (and --dynamic-retrieval skills=<on|off> flag) to enable or disable...

1.0.66-1

1.0.66-1

Added Configure subagent concurrency and depth limits in /settings (usage-based billing users) Add /chronicle skills review for reviewing proposed draft skill changes, with options to accept, reject, or defer each draft Add desktop notifications for attention prompts and idle sessions Make /share use Mission Control links for session sharing Improved Search inputs match queries that have leading or trailing whitespace Keep idle agents available after you cancel a turn Show sandbox-bypass warnings and label bypassed commands /pr auto now starts a self-paced loop that fixes one thing per run and...