feat(cli): add /output-style command with picker and mid-session switching - #10683
Conversation
…ching Bare /output-style opens a RadioButtonSelect picker (default plus the four built-in styles); /output-style <name> applies directly, case-insensitively, with 'default' selecting no style. A choice applies to the running session by rebuilding the system instruction in place and persists to user settings. general.outputStyle no longer requires a restart. Follows QwenLM#10283 as the picker/live-switch slice it deferred.
02e32b6 to
b839042
Compare
|
Please do not rebase or force-push to an active PR as it invalidates existing review comments. Note for future reference, the bots always squash all changes into a single commit automatically as part of the integration. 中文请勿对活跃的 PR 执行 rebase 或 force-push，因为这会使已有的评审评论失效。另外，供日后参考：作为集成流程的一部分，机器人始终会自动将所有改动压缩（squash）为单个提交。 |
|
Thanks for the PR! Template looks good ✓ Problem: this isn't a bug fix but the next slice in a planned series — #10283's description explicitly promised "the Direction: aligned. This is the fourth step of a merged series (#9565, #10282, #10283, all by the same author, who has write access), and the live Size: two core paths touched, both trivial — Approach: the shape feels right. It follows the established Risk: no elevated risk signals — none of the changed files match the revert-correlated paths. Moving on to code review. 🔍 中文说明感谢贡献！ 模板完整 ✓ 问题：这不是 bug 修复，而是计划系列中的下一个切片 —— #10283 的描述明确承诺了"能在会话中途切换的 方向：对齐。这是已合并系列（#9565、#10282、#10283，同一作者，具有 write 权限）的第四步，实时的 规模：触及两个核心路径，均极小 —— 方案：形态合理。沿用既有的 风险：无升级风险信号 —— 改动文件均未命中与 revert 相关的高风险路径。 进入代码审查 🔍 — Qwen Code · qwen3.8-max Reviewed at |
|
Code review My independent take before reading the diff: this wants a
No critical blockers found. One observation below the bar, for the author: the design doc justifies persisting the literal The live hand-off is the point of this PR, so here is the path a selection takes: sequenceDiagram
participant P1 as User
participant P2 as output-style command and hook
participant P3 as Config
participant P4 as LlmClient
participant P5 as LoadedSettings
P1->>P2: pick a style in the dialog, or pass it as an argument
P2->>P3: setOutputStyle(style, or undefined for default)
P2->>P4: refreshSystemInstruction()
P4-->>P2: system prompt rebuilt and rebound in place
P2->>P5: setValue(User, general.outputStyle, name or default)
P2-->>P1: report the outcome in chat
Files changed (23 of 23 shown)
Testing This is an unattended CI run: triage read the PR's own CI checks through the API and executed none of the PR's code. The tmux transcript in the PR body is the author's claim from a real Linux session, not independently re-run here. Note for the record: the commit originally pushed (
What the unit tests do pin (from the diff): every command branch, the hook's apply/clear/cancel paths, the requiresRestart flip, and the dialog's render/cancel. What they cannot pin is the behavioural heart of the PR — that the picker actually opens with the current style pre-selected, that the very next turn answers in the new style, and that non-interactive mode lists instead of opening a dialog — because those live in the rendered TUI and the model round-trip, which is exactly what the author's single Linux session demonstrates only as a claim. Sandboxed verification would settle this: 中文说明代码审查 在读 diff 之前我的独立方案是：做一个
未发现致命阻塞问题。一条低于门槛的观察，供作者参考：设计文档用"工作区级设置不会在下次启动时悄悄卷土重来"来论证清除时持久化字面量 中途切换是本 PR 的核心，选择路径如下（图：用户 → 命令/钩子：选择风格 → Config：setOutputStyle → LlmClient：refreshSystemInstruction，原地重建系统提示 → LoadedSettings：持久化到用户级 → 回到用户：在对话中反馈结果）。 测试 这是无人值守的 CI 运行：triage 通过 API 读取 PR 自身的 CI 检查，未执行任何 PR 代码。PR 正文中的 tmux 录屏是作者在真实 Linux 会话中的自述，未在此独立复跑。记录在案：最初推送的提交（ 单测钉住的内容（依据 diff）：命令的所有分支、钩子的应用/清除/取消路径、requiresRestart 翻转、对话框渲染与取消。单测钉不住的是行为核心 —— 选择器确实以当前风格预选打开、下一轮回答确实带上新风格、非交互模式列出选项而非开对话框 —— 这些活在渲染出的 TUI 和模型往返里，正是作者单平台（Linux）会话自述所覆盖、而此处无法独立核实的部分。 沙箱验证可以定论： — Qwen Code · qwen3.8-max Reviewed at |
|
Confidence: 4/5 — clean review; the code follows the repo's own precedent so faithfully that the only open items are one inaccurate design-doc sentence and CI that hasn't settled. Stepping back: this is exactly the slice #10283 carved out for separate review, and it lands the shape that PR deferred — no more, no less. The implementation is the boring version of this feature in the best sense: one shared apply path for the dialog and argument forms, reuse of the existing style machinery and the What keeps this at 4 rather than 5 is evidence, not code: the unit suite pins the command logic deterministically, but what users will actually feel — the picker opening pre-selected on the current style, the very next turn answering in the new style — rests so far on the author's single-Linux-session transcript, and CI for the reviewed head is still running. Approving now would attest to a green suite that doesn't exist yet, so approval is deferred until CI lands green on 中文说明置信度：4/5 —— 干净的审查；代码忠实沿用仓库自身先例，遗留项只有设计文档中一句不准确的表述，以及尚未结束的 CI。 退一步看：这正是 #10283 拆出来单独评审的切片，落地的也正是那个 PR 延后的形态 —— 不多不少。实现在最好的意义上是这个功能"无聊"的版本：对话框与带参形式共享一条应用路径，复用既有风格机制与 停在 4 而不是 5 的原因是证据，不是代码：单测确定性地钉住了命令逻辑，但用户真正会感受到的部分 —— 选择器以当前风格预选打开、下一轮回答带上新风格 —— 目前仍只有作者单 Linux 会话的录屏为证，且评审头的 CI 仍在运行。现在批准等于为一个尚不存在的绿色结果背书，因此批准推迟到 — Qwen Code · qwen3.8-max Reviewed at |
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Partially reviewed — gaps disclosed.
Not reviewed: build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally.
Test Plan (not a blocker): src/ui/commands/output-style-command.test.ts — no such file or directory; src/ui/hooks/use-output-style-command.test.ts — no such file or directory; src/ui/components/OutputStyleDialog.test.tsx — no such file or directory; src/config/settingsSchema.test.ts — no such file or directory.
中文说明
仅完成部分审查，审查缺口已披露。
未审查：build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally。
Test Plan（非阻断）：src/ui/commands/output-style-command.test.ts — no such file or directory; src/ui/hooks/use-output-style-command.test.ts — no such file or directory; src/ui/components/OutputStyleDialog.test.tsx — no such file or directory; src/config/settingsSchema.test.ts — no such file or directory。
— qwen3.8-max via Qwen Code /review (v0.22.3)
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Partially reviewed — gaps disclosed. Suggestions are inline.
Not reviewed: build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally.
Test Plan (not a blocker): src/ui/commands/output-style-command.test.ts — no such file or directory; src/ui/hooks/use-output-style-command.test.ts — no such file or directory; src/ui/components/OutputStyleDialog.test.tsx — no such file or directory; src/config/settingsSchema.test.ts — no such file or directory.
Deferred under the convergence posture (round 2, not a blocker) — recorded, not requested in this round:
packages/cli/src/ui/commands/output-style-command.ts:36 — [probe] supportedModes contract pinned by no test — deferred under the rounds-2-5 code-age rule (line unchanged since the previous round's reviewed head)
中文说明
仅完成部分审查，审查缺口已披露。 建议见行内评论。
未审查：build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally。
Test Plan（非阻断）：src/ui/commands/output-style-command.test.ts — no such file or directory; src/ui/hooks/use-output-style-command.test.ts — no such file or directory; src/ui/components/OutputStyleDialog.test.tsx — no such file or directory; src/config/settingsSchema.test.ts — no such file or directory。
收敛姿态下延后（第 2 轮，非阻断）——已记录，本轮不要求修改：共 1 条（原文未翻译，列表见上方英文部分）。
— qwen3.8-max via Qwen Code /review (v0.22.3)
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Not reviewed: build-and-test — unit test phase: zero suites ran in this review (install + builds exhausted the 600s per-call budget on both attempts; no recorded test scope to resume).
Not reviewed: build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally.
Test Plan (not a blocker): src/ui/commands/output-style-command.test.ts — no such file or directory; src/ui/hooks/use-output-style-command.test.ts — no such file or directory; src/ui/hooks/slashCommandProcessor.test.ts — no such file or directory; src/ui/hooks/useDialogClose.test.ts — no such file or directory; src/ui/components/OutputStyleDialog.test.tsx — no such file or directory; and 1 more.
Deferred under the convergence posture (round 3, not a blocker) — recorded, not requested in this round:
packages/cli/src/services/BuiltinCommandLoader.ts:163 — [probe] Loader registration of /output-style is untested — deleting the registration leaves every test green (272 tests measured)packages/cli/src/ui/commands/output-style-command.test.ts:27 — [probe] Test suite outcomes depend on ambient QWEN_SYSTEM_MD — 3 tests fail when the variable is exported (reproduced)
中文说明
未审查：build-and-test — unit test phase: zero suites ran in this review (install + builds exhausted the 600s per-call budget on both attempts; no recorded test scope to resume)。
未审查：build-and-test — Integration Tests (CLI, No Sandbox) was skipped in CI and its suite did not run locally。
Test Plan（非阻断）：src/ui/commands/output-style-command.test.ts — no such file or directory; src/ui/hooks/use-output-style-command.test.ts — no such file or directory; src/ui/hooks/slashCommandProcessor.test.ts — no such file or directory; src/ui/hooks/useDialogClose.test.ts — no such file or directory; src/ui/components/OutputStyleDialog.test.tsx — no such file or directory; and 1 more。
收敛姿态下延后（第 3 轮，非阻断）——已记录，本轮不要求修改：共 2 条（原文未翻译，列表见上方英文部分）。
— qwen3.8-max via Qwen Code /review (v0.22.3)
qwen-code-dev-bot
left a comment
qwen-code-dev-bot
left a comment
There was a problem hiding this comment.
Reviewed at head 360211b4.
- All three round-1 Criticals verified fixed at this head: persistence now writes the scope that actually wins (workspace-owned key keeps the user scope from being shadowed, with the untrusted-workspace write path refusing instead of half-applying), the "saved but no effect" message branches on the real cause (system-prompt replacement incl. QWEN_SYSTEM_MD vs the headless Learning skip) computed from
resolveMainSessionOutputStyle, and the description key carries entries in all three strict-parity locales. - Write failures now surface via
throwOnWriteFailureinto the dialog's error path rather than a swallowed success message; the non-interactive no-arg status reports the effective style, not the configured one. - 20/20 review threads are resolved; the picker and recording flow read clean. No new Critical issues found.
- CI on this head has no failures; per the channel convention the call is on the review itself.
yiliang114
left a comment
yiliang114
left a comment
There was a problem hiding this comment.
Review passed.
- Mid-session switching uses the existing seams end to end:
config.setOutputStyle+getLlmClient().refreshSystemInstruction()(the style lives in the stable prompt layer, so the rebuild is required and present); persistence happens BEFORE the runtime apply withthrowOnWriteFailure, so a failed write never leaves a half-applied state. - Scope handling matches the boot-time design: a trusted workspace owning
general.outputStylegets the write to workspace scope (refused when the session policy forbids workspace settings writes), otherwise user scope;--bare/--safe-moderefuse up front; non-interactive/ACP paths report the current style instead of opening a picker. - The picker follows the /effort pattern exactly (dialog registration, UIState/UIActions wiring, busy-flag inclusion, Esc cancels with no change); the dialog closes before the apply runs; cancel vs selection are distinct, and the 'saved but no effect' caveats (replaced system prompt, headless Learning skip) reuse the same resolution helper as boot.
- Unknown names are rejected with the option list;
defaultclears. Comprehensive tests for the command, utils, hook, and dialog.
CI: ubuntu Test failure is an agent-view supervisor timeout flake (worker 'did not report ready before timeout') followed by the 2h job cancel — unrelated to this PR's files; web-shell E2E passed.
chiga0
left a comment
chiga0
left a comment
There was a problem hiding this comment.
No blocking findings.
Approval blockers: none.
Checked / ran (local, linux, node v24):
- Live-switch path end to end:
setOutputStyle→LlmClient.refreshSystemInstruction→getMainSessionBaseSystemPrompt→resolveMainSessionOutputStyle(core/prompts.ts:464) →getCoreSystemPrompt; the per-turn reminder reads the same resolver, so prompt and reminder agree. - Settings writer/reader contract:
defaultpersists as the literal andresolveOutputStyle(cli config.ts:1291) maps unset/empty/defaultto no style; unknown names warn instead of locking the user out. - Scope/policy logic: trusted workspace owning the key writes back to Workspace scope, untrusted falls back to User, and
allowWorkspaceSettingsWrite: false(ACP standalone policy) refuses. The dialog path is gated onexecutionMode === 'interactive', so the restrictive non-interactive policy is not bypassed. - Sibling consistency with
/output-language(languageCommand.ts:236); bare/safe-mode guarded in both the command andapplyOutputStyleSelection. - Ran: new/changed suites 78/78 green · slashCommandProcessor 128/128 · i18n parity (mustTranslateKeys) 20/20.
- Mutation probes: setOutputStyle/refresh order swap → killed (invocationCallOrder pin) ·
settings.isTrusted &&removal → killed · zh parity translation removal → killed · command-level bare/safe guard removal → survives as an equivalent mutant (the utils-layer guard backstops with an identical refusal; wording-only delta).
Cross-check vs prior ci-bot findings at this head: R1-1/R1-2/R1-3 criticals fixed (workspace-owns scope + isTrusted; correct no-effect reason; zh/zh-TW parity restored) · R1-4's four unpinned behaviours now pinned · R1-5 requiresRestart restored to true · R1-6 write failure surfaced via throwOnWriteFailure · R1-7 no-arg message reports the effective style · R1-8 bare/safe guards + docs row · R1-9 design doc dropped from the PR · R2-1..R2-7 gaps closed in the final commit. Verified by reading at head and by the suites above.
Not covered: the repo-wide unit suite beyond the changed surface · the integration-test legs · Windows/macOS terminal behaviour (TUI-only change, linux evidence).
Context on the CI red behind the latest bot downgrade: Test (ubuntu-latest, Node 22.x) shows one failing test, src/agent-view/supervisor-process.test.ts, in a package this PR does not touch, before the job was cancelled; not attributed to this change.
Reviewed with AI assistance.
…output-styles A style is a Markdown file whose body is the prompt section, with an optional frontmatter (name, description, keep-coding-instructions) that defaults to the file name, the first body line, and false. The catalog is built-ins plus the user's files plus, in a trusted workspace, the project's, with project > user > built-in precedence by case-insensitive name. The startup resolver, /output-style <name>, and the picker all read the catalog, so a new file needs no restart. --bare and --safe-mode keep the built-ins only; invalid files are skipped and logged rather than fatal. Follows QwenLM#10683 as the custom-style slice it deferred.
|
Released in v0.23.0. |
…output-styles (QwenLM#10761) * feat: load custom output styles from ~/.qwen/output-styles and .qwen/output-styles A style is a Markdown file whose body is the prompt section, with an optional frontmatter (name, description, keep-coding-instructions) that defaults to the file name, the first body line, and false. The catalog is built-ins plus the user's files plus, in a trusted workspace, the project's, with project > user > built-in precedence by case-insensitive name. The startup resolver, /output-style <name>, and the picker all read the catalog, so a new file needs no restart. --bare and --safe-mode keep the built-ins only; invalid files are skipped and logged rather than fatal. Follows QwenLM#10683 as the custom-style slice it deferred. * fix(output-style): stop a style file reading anything but itself Two ways a style file could carry content it was never meant to. A style file's body goes into the system prompt verbatim, and the loader followed links with only an `fs.stat`. A repo can commit `.qwen/output-styles/notes.md -> ~/.aws/credentials` plus a `.qwen/settings.json` naming it; folder trust is off by default, so `git clone && qwen` treats the workspace as trusted, `stat.isFile()` reports on the *target*, and the secret is embedded under `# Output Style: notes` with no tool call and no approval. A hard link (`nlink > 1`) and a symlinked ancestor (a checked-in `.qwen -> /outside`, which a final-component check cannot see) reach the same place. This mirrors `readLoopTaskFile`, which guards the identical sink. A project file is read with `lstat`, so a link is not a regular file and never reaches the read -- confinement alone would wave through `notes.md -> .env`, whose target is inside the root. A user file may be a link, because `~/.qwen/output-styles/x.md -> ~/dotfiles/x.md` is an ordinary setup, but its canonical path must stay inside the home root. Both refuse `nlink > 1` and both confine the realpath. Separately, the description reached the terminal un-sanitized while the sibling `name` from the same untrusted file was refused for exactly those characters, and a declared description had no length bound: an OSC-8 hyperlink made a picker row link somewhere it did not say, and U+202E reordered the row so the `(project)` marker read as something else. Both description sources now go through `stripAnsiAndControl` plus a `\p{Cf}` strip, collapse to one line, and share the cap the derived description already used. Each guard is pinned by a case that goes red without it, including the in-workspace symlink that distinguishes `lstat` from confinement. * fix(output-style): key the headless Learning rule on the built-in, not the name The docs this PR adds tell users a file "overrides a built-in of that name", so a user customizing Learning ships `~/.qwen/output-styles/Learning.md`. The lowercase dedupe key shadows the built-in and the name keeps the basename's capital L, so `resolveEffectiveOutputStyle`'s `name === 'Learning'` check fires on the user's own prompt: `qwen -p` then carries no output style at all -- no prompt section, no turn reminder -- and `warnAboutOutputStyle` stays quiet because the name resolved, so it is a silent no-style run. Renaming the file `learning.md` shadows the identical built-in and *is* applied, so two spellings of one name behaved differently in a module that dedupes and looks names up case-insensitively everywhere else. The rule exists because the built-in Learning prompt tells the model to stop and wait for a reply a headless run cannot send. A custom file that happens to take the name carries no such instruction, so the check keys on `source === 'built-in'` as well. * fix(output-style): stop a dismissed picker re-opening behind the user `openOutputStyleDialog` suspends on `loadSessionOutputStyles` and then sets the dialog open. A dismissal landing in that window leaves the continuation queued, so the dialog the user just closed re-opens and the next Enter is captured by `RadioButtonSelect`, which writes `general.outputStyle` with `{ throwOnWriteFailure: true }`. An open generation, retired both by a newer open and by a selection, makes the stale continuation a no-op. The defect had no in-suite witness because this diff deleted `act(() => result.current.openOutputStyleDialog())` from `clears the style when "default" is chosen` rather than adapting it with the siblings' `waitFor`. Its closing `expect(isOutputStyleDialogOpen).toBe(false)` then read the `useState(false)` initial value and passed for any implementation, including one that never closes. The open is restored with the wait, and a new case dismisses an open and then resolves its load, asserting the dialog stays closed. * fix(opentui): give the output-style picker the same catalog as ink Startup style resolution is renderer-independent, so a custom style can be active under OpenTUI via `--output-style`, `general.outputStyle`, or the renderer-agnostic `/output-style <name>`. The dialog built its list from `BUILT_IN_OUTPUT_STYLES` alone, so the active style was not in it, `findIndex` returned `-1`, and `Math.max(0, -1)` highlighted `default` -- contradicting the comment two lines below claiming index 0 tells the truth. One Enter then applied `undefined` and persisted `"outputStyle": "default"` with `{ throwOnWriteFailure: true }`, clearing the user's own setting. The docs line this PR adds, that custom styles appear in the picker labelled with their source, was false on this renderer. The dialog now loads through `loadSessionOutputStyles`, which carries the ink path's trust gate, and labels non-built-ins with their source the way `describe()` does. `useState` runs its initialiser once, so the pre-selection is re-derived when the catalog arrives -- otherwise the `-1 -> 0` clamp would survive the fix. The test stubs the catalog load so it never reads the developer's own `~/.qwen/output-styles`, and asserts a `user` style renders as a selectable row and is the pre-selected one. * fix(output-style): make picker loading non-interactive * fix(opentui): keep the output-style picker steady while it is open The catalog effect depended on the `onClose` and `notify` props, which the mount site rebuilds as inline closures on every render, so any shell re-render tore down the settled load and re-read both style directories. The reload landed a new array, the selection was re-derived, and the user's arrow-key navigation was discarded -- Enter then applied the previously active style. Keep the callbacks in refs and depend only on `config`. The catalog is also re-read on every open and skips a file it cannot parse, so the active style can be missing from it while the session still runs it. Snapping to index 0 marked `default` as the active style and one Enter persisted that over the user's setting; list the live definition instead, matching membership case-insensitively the way the catalog dedupes. * fix(output-style): read a style file the way a prompt should be read The loader took plausible Markdown files and turned them into something the author did not write. A `---` rule around prose was parsed as frontmatter and its text vanished from the prompt, while a fence with a trailing space was not frontmatter at all and its raw YAML became the prompt; an HTML comment meant for a teammate was fed to the model, and a comment-only template file loaded as a selectable style; a UTF-16 file -- what PowerShell's `>` and Notepad write -- decoded to NUL-riddled mojibake and still reached the system prompt; the size bound was 41x the house bound for this sink and was checked against a `stat` the file could outgrow between the check and the read. A file that shadows a built-in also inherited `keep-coding-instructions: false`, so customizing the wording of `concise.md` silently deleted the verification and faithful-reporting guidance, and a declared `True` was read as false whenever an unrelated frontmatter line made the YAML parser fall back to its simple mode. Alongside those: the file-name-derived name is trimmed, so a stray trailing space no longer makes an entry that is listed but unselectable and ` default.md` no longer bypasses the reserved name; a non-string YAML `name` falls back to the file name instead of throwing an opaque TypeError or loading as `a,b`; and the home-directory guard compares the directories that would actually be read, so a relocated `QWEN_HOME` neither hides the project level nor relabels the user's own files. `stripHtmlComments` moves to `textUtils` next to `normalizeContent`, the home the rules loader's copy should always have had. * fix(output-style): keep a resolved style inside the session it belongs to Two consumers took the resolved style further than it was meant to go. Project trust was enforced only where the catalog is read, so a checked-in style kept shaping every turn after the workspace lost trust mid-session -- the IDE companion changes that verdict in place, with no restart -- while the picker had already stopped listing it. The gate now sits in the resolver the prompt and the per-turn reminder share, so both follow the flip. Arena peers were handed `getOutputStyle()` raw, so a peer inherited a style even when the main session's prompt was replaced, and a style with `keepCodingInstructions: false` stripped the software-engineering guidance -- verification and faithful reporting included -- from agents whose whole job is to produce a diff. The trust mock in the config tests answered the same way whatever it was asked, and the unknown-style warning's "Available styles" list had no test at all; both are now pinned. * fix(output-style): say what the setting takes, and where a choice lands Four surfaces described the value `--output-style` and `general.outputStyle` accept: two still listed only the four built-ins, and two named the style *file* where the code matches the declared frontmatter `name` -- so a user with `review-v2.md` declaring `name: Reviewer` typed `review-v2`, following the editor hover attached to the very file being edited, and got an unknown-style warning on every startup. Selecting a style also wrote the name wherever the key already lived, without saying so: a project style's name into user settings, where no other project can resolve it, or a personal style's name into the shared checked-in workspace file, where every teammate and CI run warns and runs unstyled. The scope stays where it is -- moving it would be shadowed by the workspace value -- and the confirmation now names the mismatch. The ink picker matched the active style name exactly while every other lookup is case-insensitive, and an active style the re-read catalog no longer carries fell back to `default`, where one Enter persisted it over the user's setting. Also restores the design doc and the E2E plan dropped by an earlier fix commit, with the corrections their reviews asked for: the picker resolves against the open-time snapshot, so a file deleted after the picker opened still applies and its persisted name warns at the next startup -- the opposite of what the bullet claimed -- and the plan's baseline, width, trust setup and restore steps are now executable and falsifiable. * fix(cli): explain inactive project output styles --------- Co-authored-by: 易良 <1204183885@qq.com> Co-authored-by: qqqys <266654365+qqqys@users.noreply.github.com> Co-authored-by: Shaojin Wen <shaojin.wensj@alibaba-inc.com>
5 participants
What this PR does
This adds
/output-styleso users can inspect, choose, and change the active output style from inside a session. Bare/output-styleopens a picker withdefaultand the four built-in styles, pre-selected on the effective style;/output-style <name>applies a style case-insensitively; and/output-style defaultclears it. In non-interactive and ACP sessions, the bare form reports the current style and options instead of opening a dialog.A successful choice persists first: it writes a trusted workspace when that workspace owns
general.outputStyle; otherwise it writes user settings. After that write succeeds, the command updates the live configuration and rebuilds the bound system instruction so the next turn uses the new style. Clearing persists the literaldefaultin that owning scope. When a custom system prompt replaces the built-in prompt, or Learning is skipped in a headless run, the command saves the choice and reports the actual effective behavior. Bare and safe modes reject changes, and persistence failures leave the running style unchanged.The generic settings path still marks
general.outputStyleas requiring a restart because only/output-styleperforms the live refresh. The user documentation now describes the built-in styles and both configuration paths.Why it's needed
#9565 shipped the styles and #10283 made them selectable at startup, but users still could not discover or change the effective style inside a running session. This closes that gap with the established
/effortcommand shape while keeping the live refresh and persistence behavior explicit and testable.Reviewer Test Plan
How to verify
Run
npm run dev, then invoke/output-style. Confirm the picker opens with the effective style selected, choosingConcisereports success and changes the next response, and Escape or the dialog-close path cancels without mutation. Confirm/output-style conciseapplies case-insensitively,/output-style defaultclears and persistsdefault, and an unknown name reports the valid choices without changing runtime or settings. In-por ACP mode, confirm the bare command reports the current style and options instead of opening the picker.For scope behavior, confirm a trusted workspace that owns the key is updated in workspace scope, while an untrusted or non-owning workspace writes user scope. Confirm bare and safe modes reject the command, a persistence error does not change the live style, and a live refresh failure still reports the persisted result.
Focused verification from
packages/cli:Result: 6 files passed, 206 tests passed. The ordering assertion was also mutation-checked: moving the live config update after system-instruction refresh makes both set and clear tests fail.
Evidence (Before & After)
Before:
/output-stylewas not a command; the active style was invisible and changing it required editing settings and restarting.After:
Tested on
Environment (optional)
Node 22;
npm run devfor the interactive session; focused Vitest coverage plus repository build for review fixes.Risk & Scope
.qwen/output-styles/*.mdfiles, extension-provided styles, and a footer/status-line indicator are later slices./output-styleis the dedicated live-update path.Linked Issues
Follows #9565, #10282, and #10283.
中文说明
这个 PR 做了什么
本 PR 新增
/output-style，让用户能在会话内查看、选择和更改当前输出风格。裸/output-style打开选择器，列出default和四个内置风格，并预选当前实际生效的风格；/output-style <name>以大小写不敏感的方式直接应用；/output-style default清除风格。在非交互和 ACP 会话中，裸命令报告当前风格和可选项，而不是打开对话框。选择成功时先持久化：若可信工作区拥有
general.outputStyle，则写回工作区级设置；否则写入用户级设置。写入成功后，命令再更新当前运行配置并重建已绑定的系统指令，因此下一轮即可使用新风格。清除操作会在对应的拥有者作用域持久化字面量default。当自定义系统提示词替换内置提示词，或 Learning 在 headless 模式下被跳过时，命令仍保存选择，并准确报告实际生效情况。Bare 和 safe 模式拒绝更改；持久化失败时不会改变运行中的风格。通用设置路径仍将
general.outputStyle标记为需要重启，因为只有/output-style会执行实时刷新。用户文档现已说明内置风格和两种配置路径。为什么需要它
#9565 提供了输出风格，#10283 让启动时可以选择风格，但用户仍无法在运行中的会话里发现或更改实际风格。本 PR 使用既有的
/effort命令形态补齐该能力，并让实时刷新和持久化行为都有明确测试覆盖。评审者测试计划
如何验证
运行
npm run dev后调用/output-style。确认选择器打开并预选实际生效的风格；选择Concise后报告成功，下一轮回答采用新风格；Escape 或关闭对话框不会产生变更。确认/output-style concise大小写不敏感，/output-style default清除并持久化default，未知名称会列出有效选项且不改变运行时或设置。在-p或 ACP 模式下，裸命令应报告当前风格和可选项，而不是打开选择器。作用域方面，确认拥有该键的可信工作区写回工作区级设置，不可信或不拥有该键的工作区写入用户级设置。确认 bare/safe 模式拒绝命令，持久化错误不会改变实时风格，实时刷新失败时仍报告已持久化的结果。
在
packages/cli下执行上方 focused Vitest 命令：6 个文件、206 项测试全部通过。调用顺序还做了反向变异验证：将运行配置更新故意移动到系统指令刷新之后，两条顺序测试都会失败。证据（Before & After）
Before：
/output-style不存在；当前风格不可见，更改风格需要编辑设置并重启。After：见上方真实会话记录。会话中选择 Concise 后，紧接着的回答已经采用“结果先行、零铺垫”的风格；未知风格报错且不修改状态；
default恢复默认风格。测试环境
风险与影响范围
.qwen/output-styles/*.md文件、extension 提供的风格、footer/状态栏指示属于后续切片。/output-style是专用的实时更新路径。关联 Issue
承接 #9565、#10282 与 #10283。