flower-trellis 0.2.2 → 0.2.4

package/enhancements/0.6/.agents/skills/trellis-push/SKILL.md

@@ -72,6 +72,8 @@ git status --short
 
 对每个有变更的仓库，依次执行以下操作：
 
+> **commit-only 模式**（调用方传入「只提交不推」/「commit-only」语义，如 Trellis Phase 3.4「commit now, push later」）：只执行 2.1 展示 → 2.2 暂存确认 → 2.3 commit，**跳过 2.4 push 与 2.5 merge**。Step 3 快照、Step 4 结果照常，结果中标注「本地已提交、未推送，后续仍走 trellis-push 推」。
+
 #### 2.1 展示变更摘要
 
 ```bash
@@ -105,6 +107,8 @@ git commit -m "<type>(<scope>): <description>"
 
 #### 2.4 Push 当前分支
 
+> **commit-only 模式:跳过 2.4 与 2.5**，直接进 Step 3。
+
 ```bash
 git push origin <current_branch>
 ```
@@ -291,6 +295,7 @@ git remote -v | grep -E "^origin\s+"
 | 语义 | 说明 | 用户怎么说 |
 |------|------|-----------|
 | 默认 | 自动检测所有有变更的仓库并处理 | `/trellis-push` |
+| commit-only | 只 commit 不 push（Phase 3.4「commit now, push later」），跳过 2.4/2.5 | 「只提交不推」/「commit-only」 |
 | 指定仓库 | 只处理指定仓库 | 「只 push 前端」/「push frontend」 |
 | 重新配置 | 重新询问目标分支 | 「重新配置 push 目标分支」/「reconfigure push」 |
 | 临时目标 | 临时指定目标分支（不修改配置） | 「push 到 hotfix 分支」 |

package/enhancements/0.6/.claude/skills/trellis-push/SKILL.md

@@ -72,6 +72,8 @@ git status --short
 
 对每个有变更的仓库，依次执行以下操作：
 
+> **commit-only 模式**（调用方传入「只提交不推」/「commit-only」语义，如 Trellis Phase 3.4「commit now, push later」）：只执行 2.1 展示 → 2.2 暂存确认 → 2.3 commit，**跳过 2.4 push 与 2.5 merge**。Step 3 快照、Step 4 结果照常，结果中标注「本地已提交、未推送，后续仍走 trellis-push 推」。
+
 #### 2.1 展示变更摘要
 
 ```bash
@@ -105,6 +107,8 @@ git commit -m "<type>(<scope>): <description>"
 
 #### 2.4 Push 当前分支
 
+> **commit-only 模式:跳过 2.4 与 2.5**，直接进 Step 3。
+
 ```bash
 git push origin <current_branch>
 ```
@@ -291,6 +295,7 @@ git remote -v | grep -E "^origin\s+"
 | 语义 | 说明 | 用户怎么说 |
 |------|------|-----------|
 | 默认 | 自动检测所有有变更的仓库并处理 | `/trellis-push` |
+| commit-only | 只 commit 不 push（Phase 3.4「commit now, push later」），跳过 2.4/2.5 | 「只提交不推」/「commit-only」 |
 | 指定仓库 | 只处理指定仓库 | 「只 push 前端」/「push frontend」 |
 | 重新配置 | 重新询问目标分支 | 「重新配置 push 目标分支」/「reconfigure push」 |
 | 临时目标 | 临时指定目标分支（不修改配置） | 「push 到 hotfix 分支」 |

package/enhancements/0.6/overrides/workflow-states/in_progress-inline.md

@@ -1,5 +1,8 @@
 <!-- BEGIN skill-garden workflow-state in_progress_inline v0.6 -->
 HIGHEST PRIORITY SKILL-GARDEN STATE GUARD (in_progress-inline):
 Inline mode means the main session edits directly, but check still routes before `trellis-check` / `trellis-check-all`.
+After `trellis-check` / `trellis-check-all`, stop and report; do not run `/trellis:finish-work` unless the user explicitly asks after Phase 3.4 is complete.
+This guard overrides any lower `Flow: ... -> /trellis:finish-work` line in this state block.
+At Phase 3.4, code commit/push still goes through `trellis-push` (commit-only for commit-without-push); never bare `git commit`/`git push` on code (hub: Code Commit Confirmation Gate).
 If active task.json has `last_push_snapshot`, relay `partial_step` + `next_step` once before starting new work.
 <!-- END skill-garden workflow-state in_progress_inline v0.6 -->

package/enhancements/0.6/overrides/workflow-states/in_progress.md

@@ -3,5 +3,8 @@ HIGHEST PRIORITY SKILL-GARDEN STATE GUARD (in_progress):
 At Phase 2.1/2.2/3.1, invoke `trellis-route(implement|check)` first, including every check / check-all path.
 Do not spawn `trellis-implement`, `trellis-check`, or `trellis-check-all` directly unless `trellis-route` just selected subagent mode.
 If routing helper is unavailable, ask the same numbered route choices in normal chat and wait for the user's selection.
+After `trellis-check` / `trellis-check-all`, stop and report; do not run `/trellis:finish-work` unless the user explicitly asks after Phase 3.4 is complete.
+This guard overrides any lower `Flow: ... -> /trellis:finish-work` line in this state block.
+At Phase 3.4, code commit/push goes through `trellis-push` (commit-only mode for commit-without-push); never bare `git commit`/`git push` on code (hub: Code Commit Confirmation Gate).
 If active task.json has `last_push_snapshot`, relay `partial_step` + `next_step` once before starting new work.
 <!-- END skill-garden workflow-state in_progress v0.6 -->

package/enhancements/0.6/overrides/workflow.md

@@ -6,7 +6,7 @@
 
 **Priority**: This hub overrides any conflicting Trellis workflow, skill, or command text for the scoped behaviors below.
 
-**Scope**: Phase 2.1 / 2.2 / 3.1 dispatch routing, Phase 3.5 finish-work bookkeeping, and push-progress recovery / snapshot reminders. State blocks should keep only one short skill-garden sentinel per state; long-form rules live here.
+**Scope**: Phase 2.1 / 2.2 / 3.1 dispatch routing, post-check stop, Phase 3.4 code commit/push via trellis-push, explicit Phase 3.5 finish-work bookkeeping, and push-progress recovery. State blocks should keep only one short skill-garden sentinel per state; long-form rules live here.
 
 **Mechanical rule**: use this hub as the source of truth. Do not add separate top-level skill-garden override sections or multiple skill-garden sentinels inside the same `workflow-state:*` block.
 
@@ -14,7 +14,7 @@
 
 Before any implement/check agent or check skill runs from the main session, the immediately preceding routing decision must come from `trellis-route` or from the same numbered fallback choices shown in normal chat when the helper is unavailable.
 
-`trellis-route` returns 4 modes for `target=check` (check-all/check x inline/subagent) and 2 for `target=implement`. Step 1.7's recommendation is generated inside the skill and surfaced via Step 2's `AskUserQuestion`; that is the only prompt point.
+`trellis-route` returns 4 modes for `target=check` (check-all/check x inline/subagent) and 2 for `target=implement`. The route prompt is the only place to ask this choice.
 
 If the platform cannot call `AskUserQuestion` / `request_user_input`, ask the same numbered choices in normal chat and wait for the user's reply. Tool unavailability is not permission to record inline/subagent or to dispatch a sub-agent directly.
 
@@ -27,22 +27,23 @@ At phase boundaries, do not ask meta continuation questions such as "continue?",
 
 Check routing has no 4h preference file. Before `trellis-check`, `trellis-check-all`, or either check sub-agent, route every time so the user can choose check-all vs lightweight and inline vs subagent.
 
-#### Finish-work Bookkeeping Guard
+#### Post-Check Stop Gate
 
-`session_auto_commit` governs ONLY the script-managed bookkeeping commits that `task.py archive` and `add_session.py` make for their own files (`.trellis/tasks/**` archive moves and `.trellis/workspace/**` journal/index). It is NOT a global auto-commit switch, and it has NO authority over code commits.
+After `trellis-check` or `trellis-check-all` finishes, stop and report the result. Do not run `/trellis:finish-work`, do not archive the task, and do not imply the task is ready to wrap up solely because checks passed.
 
-Code commits always belong to Phase 3.4: the agent drafts the batched commit plan and commits only after the user confirms — in EVERY case, whether `session_auto_commit` is `true` or `false`. Never treat `session_auto_commit: true` as permission to auto-commit code, to skip the Phase 3.4 confirmation prompt, or to commit any path outside the two scripts' own bookkeeping files. The two are independent: this switch never decides whether code is committed, and it never decides whether to ask first.
+If checks pass, the next allowed workflow steps are Phase 3.3 `trellis-update-spec` and Phase 3.4 `trellis-push`/commit confirmation. `/trellis:finish-work` is explicit-only: run it only after Phase 3.4 is complete and the user asks to wrap up, archive, or finish the task.
 
-When `session_auto_commit: true` (the default): `task.py archive` produces a `chore(task): archive ...` commit and `add_session.py` produces a `chore: record journal` commit, each touching only its own bookkeeping files. Code stays dirty for the Phase 3.4 plan.
+#### Code Commit Confirmation Gate
 
-When `.trellis/config.yaml` sets `session_auto_commit: false`, finish-work must treat archive and journal writes as disk-only bookkeeping:
+Code commit/push belongs only to Phase 3.4 and must go through `trellis-push`; the main session must not run bare `git commit` / `git push` for code.
 
-- Running `python3 ./.trellis/scripts/task.py archive <task>` may move task files, but must not be described as producing a `chore(task): archive ...` commit.
-- Running `python3 ./.trellis/scripts/add_session.py ...` may write workspace journal/index files, but must not be described as producing a `chore: record journal` commit.
-- Do not run a compensating `git add` / `git commit` for `.trellis/tasks/**` or `.trellis/workspace/**` just because those scripts skipped auto-commit.
-- Report the resulting `.trellis/tasks/**` and `.trellis/workspace/**` dirty paths to the user as bookkeeping changes for manual review.
+`trellis-push` confirmation must show both the exact file list to stage and the drafted commit message. Before the user approves that concrete list + message, do not `git add`, commit, or push; never use `git add -A` / `git add .`.
 
-This guard scopes `session_auto_commit` to the two bookkeeping commits above; Phase 3.4 code-work commits stay confirmation-gated regardless of its value.
+For "commit now, push later", use `trellis-push` commit-only mode; the later push still goes through `trellis-push`. `session_auto_commit` never authorizes code commits; it only affects bookkeeping commits below.
+
+#### Bookkeeping Auto-commit Scope
+
+`session_auto_commit` only governs the bookkeeping commits `task.py archive` / `add_session.py` make for their own `.trellis/tasks/**` and `.trellis/workspace/**` files — never code (gated above). When `false`, those archive/journal writes stay disk-only (no compensating `git commit`).
 
 #### Push Progress Recovery / Snapshot
 

package/enhancements/MANIFEST.json

@@ -1,7 +1,7 @@
 {
-  "syncedAt": "2026-06-08T17:00:12.577Z",
+  "syncedAt": "2026-06-09T03:17:57.676Z",
   "syncedFrom": "vendor/skill-garden/.trellis",
-  "sourceCommit": "1dcf26869e7c90be0777ebe3448501f340e05b45",
+  "sourceCommit": "c027d5baf4601edfe50791c5f2f34eaf41bb1357",
   "variants": {
     "old": {
       "claudeSkills": [],

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "flower-trellis",
-  "version": "0.2.2",
+  "version": "0.2.4",
   "description": "一键安装/升级 Trellis 并自动融合 skill-garden 强化包(默认 Claude + agents)",
   "type": "module",
   "bin": {
@@ -12,10 +12,10 @@
     "node": ">=18.17.0"
   },
   "dependencies": {
+    "@inquirer/prompts": "^8.5.2",
     "@mindfoldhq/trellis": "0.6.0-beta.8",
     "chalk": "^5.6.2",
     "figlet": "^1.11.0",
-    "inquirer": "^9.3.8",
     "node-pty": "^1.1.0"
   },
   "files": [

package/src/cli.js

@@ -2,6 +2,7 @@ import fs from "node:fs";
 import path from "node:path";
 import { flowerVersion, trellisVersion } from "./lib/versions.js";
 import { selectVariant } from "./lib/variant.js";
+import { readManifest } from "./lib/manifest.js";
 import { runTrellis } from "./lib/trellis-runner.js";
 
 /**
@@ -21,6 +22,10 @@ function printVersion(cwd) {
       const { version } = selectVariant(cwd);
       if (version) console.log(`project .trellis  ${version}`);
     }
+    // 项目里 flower 上次铺包时戳入的自身版本(来自 .flower-manifest.json);
+    // 与首行「当前工具版本」对比即可看出该项目是否需要重新 update。
+    const mf = readManifest(cwd);
+    if (mf && mf.flowerVersion) console.log(`project flower    ${mf.flowerVersion}`);
   } catch {
     // 忽略:版本读取失败不应影响 -v 输出
   }

package/src/lib/apply-enhancements.js

@@ -7,6 +7,7 @@ import { copySkills } from "./copy-skills.js";
 import { injectWorkflow } from "./workflow-inject.js";
 import { applyCodexTweaks } from "./codex-tweaks.js";
 import { readManifest, writeManifest } from "./manifest.js";
+import { flowerVersion } from "./versions.js";
 import { rmrf } from "./fs-utils.js";
 
 /** 清理升级后可能变空的强化目录(深 → 浅)。 */
@@ -101,7 +102,15 @@ export function applyEnhancements(target, opts = {}) {
         pruneEmptyDirs(target);
       }
     }
-    writeManifest(target, { variant, version, skills: installed, paths: newPaths });
+    // 写 manifest 时同时戳入 flower-trellis 自身版本(flowerVersion),与 version(项目 Trellis
+    // 版本)区分开:前者答「上次是哪个 flower 铺的包」,服务后续升级/维护判断。
+    writeManifest(target, {
+      flowerVersion: flowerVersion(),
+      variant,
+      version,
+      skills: installed,
+      paths: newPaths,
+    });
   }
 
   // workflow 注入:无过滤名(全装)或显式指定 workflow-enhancement/finish-work-enhancement 时执行

package/src/lib/pick-platforms.js

@@ -1,4 +1,4 @@
-import inquirer from "inquirer";
+import { checkbox } from "@inquirer/prompts";
 
 /**
  * 平台多选清单 —— flower 自己的平台选择菜单。
@@ -39,19 +39,21 @@ const PLATFORMS = [
  */
 export async function pickPlatforms() {
   if (!process.stdin.isTTY) return ["--codex", "--claude"];
-  const { tools } = await inquirer.prompt([
-    {
-      type: "checkbox",
-      name: "tools",
-      message:
-        "选择要配置的 AI 工具(空格勾选 / 回车确认,默认已勾 Claude Code + Codex):",
-      choices: PLATFORMS.map((p) => ({
-        name: p.name,
-        value: p.value,
-        checked: !!p.checked,
-      })),
-      loop: false,
-    },
-  ]);
+  // 用 @inquirer/checkbox(现代 @inquirer/core 增量重绘内核)替代经典 inquirer。
+  // 为什么换:经典 inquirer 每次重绘都「整块清屏 → 整块重写」,在 WSL2 / ConPTY 终端下
+  // 清空与重画之间有一帧空白,上下切换平台时肉眼可见闪屏;新内核按差异增量重绘可消闪。
+  // pageSize 取平台总数,一屏展示全部、避免滚动带来的二次重绘;loop:false 保持首尾不循环。
+  // 返回值即选中的 value 数组(["--claude","--codex", ...]),对调用方契约不变。
+  const tools = await checkbox({
+    message:
+      "选择要配置的 AI 工具(空格勾选 / 回车确认,默认已勾 Claude Code + Codex):",
+    choices: PLATFORMS.map((p) => ({
+      name: p.name,
+      value: p.value,
+      checked: !!p.checked,
+    })),
+    loop: false,
+    pageSize: PLATFORMS.length,
+  });
   return tools;
 }

package/src/lib/update-check.js

@@ -1,6 +1,6 @@
 import { spawnSync } from "node:child_process";
 import { fileURLToPath } from "node:url";
-import inquirer from "inquirer";
+import { confirm } from "@inquirer/prompts";
 import chalk from "chalk";
 import { flowerVersion } from "./versions.js";
 
@@ -95,7 +95,7 @@ export function isRunningViaNpx() {
  * 网络探测失败、已是最新或本地更高。
  *
  * 发现新版时的行为:
- *  - 交互 TTY:打印通知 → inquirer 询问是否升级 → 同意则执行 `npm i -g flower-trellis@latest`;
+ *  - 交互 TTY:打印通知 → confirm 询问是否升级 → 同意则执行 `npm i -g flower-trellis@latest`;
  *    成功后打印「请重新运行」并 **process.exit(0)**(不做 re-exec 自动重跑),失败则降级为
  *    打印手动升级命令并继续主流程;拒绝则继续主流程。
  *  - 非交互(`-y`/`--yes` 或非 TTY):仅打印通知 + 升级命令,不弹确认、不阻塞。
@@ -136,15 +136,11 @@ export async function checkForUpdate(ctx, commandLabel) {
     return;
   }
 
-  // 7. 交互:询问是否升级
-  const { doUpgrade } = await inquirer.prompt([
-    {
-      type: "confirm",
-      name: "doUpgrade",
-      message: `是否现在升级到 ${latest}?(升级后需重新运行命令)`,
-      default: true,
-    },
-  ]);
+  // 7. 交互:询问是否升级(@inquirer/confirm,返回 boolean)
+  const doUpgrade = await confirm({
+    message: `是否现在升级到 ${latest}?(升级后需重新运行命令)`,
+    default: true,
+  });
   if (!doUpgrade) {
     console.log("  · 已跳过升级");
     return;