comate-cli 0.12.3__tar.gz → 0.12.4b1__tar.gz

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/CHANGELOG.md

@@ -4,6 +4,18 @@
 
 ### Added
 
+- 输入框裸词 `ultracode`：普通消息里按空白切出的整词 `ultracode`（不分大小写；紧贴标点 / 引号 /
+  括号、带空格写法都不算）在输入框里显示为深紫流光（与 `/effort` 刻度尺 ultracode 格同色），发送那一刻等价于 `/effort ultracode`
+  但只改本会话、不写 settings（忙时也立即翻开关）；发行版配了 `ultracode_notice` 时先弹确认框，
+  取消则消息留在输入框；这个词保留在发给模型的消息里，scrollback 回显静态染紫。
+- ultracode 的 SDK 侧含义升级为「Workflow 激活点 + 不计成本的探索发现」：开着时 Workflow 预算放宽，
+  内层 agent 带「宁可多读，不要早停」前言，探索产出约定落 `.scratch/<feature>/research/`。
+- Workflow 预算分两档：普通档收紧为并发 4 / 每跑 20 次 agent 调用（有意克制，教材 Scale 表
+  随之改为 ≤5 / ≤10 / ≤16），ultracode 档放宽为并发 10 / 每跑 100 次；单 agent 超时两档都不变。
+- Workflow 唤起改为提示词约束：ultracode 关着时模型不主动起 workflow，只在用户用自己的话要了
+  （如 workflow / 工作流 / 编排 / 动态工作流）或明确同意模型的建议时才起；代码层不再按用户文字
+  判定拒绝，只靠预算上限兜底。
+
 - assistant 正文刷入终端改为按行节拍：整行按固定间隔进入底部暂存窗，不再模型吐多快
   屏幕刷多快。默认每行 500 毫秒；空行不占间隔；积压越多放得越快（每 20 行加一倍速，最快 4 倍），
   永远不一次性倒出；
@@ -19,6 +31,15 @@
 
 ### Changed
 
+- 启动 logo 固定 1x：不再按终端宽度自动放大。宽终端原来的 8 行 2x 大字标改回 4 行小字标（4 行排版契约不变），2x 放大相关代码随之删除；测试锁定「宽终端也必须是 1x」防回归。白牌发行版（hico）同款改动。
+- 首启向导（Provider setup）整体重做，照 `/effort` 刻度尺模式改为自绘组件 `preflight_wizard_ui.WizardScreenUI`，不再借用共用的 `SelectionMenuUI`：六屏统一「标题 / 空行 / 内容 / 空行 / 键位提示」顶部对齐骨架；选项行改为 `❯ label` 无圆点、项间不留空行；provider 屏只在选中项下方显示一行 preset 说明；API key / endpoint / 模型名输入屏去掉边框；确认屏的 JSON 预览改为对齐缩进的静态块；键位提示只写真正绑定的键（去掉并不存在的 k/j）。其它菜单不受影响。
+- `/effort` 的 `ultracode` 改绑 `max` 档（原 `xhigh`）：选 ultracode = 最高思考档 + workflows
+  多 agent 编排；`/effort max` 或 `auto` 关开关，`xhigh` 变回普通档。刻度尺描述改为
+  「max + workflows」。
+- 不支持 effort 的模型上开 ultracode，结果行不再只说「当前模型不支持 effort，设置已保存但
+  不会发送」——workflows 那一半其实已经开了。现在分开说「ultracode 已开启」和「max 已保存
+  但不会发送」；刻度尺 ultracode 格附注同步改为「思考力度不生效，仅 workflows」，
+  不再套用普通档位的「该模型不支持」。
 - 新版本提示统一指向 `comate update`：macOS/Linux 的启动一行提示与全屏弹窗「立即更新」
   标签不再展示 `uv tool install …` 长串（与 Windows 对齐），措辞改为「退出后运行」；
   更新失败的兜底文案仍保留完整 uv 修复命令。源码 editable 安装下自更新必然被拒，
@@ -26,6 +47,7 @@
 
 ### Fixed
 
+- 首启向导在矮终端里 provider 列表被裁掉、光标移到看不见的项上仍可回车的问题：旧实现写死 `page_size=10`（公版已有 12 个 preset）且选项窗不滚动。现在内容区按终端高度自适应、按行滚动、选中项永远可见，`Showing a-b/n` 只报真正画出来的项；极矮终端按「空行 → 提示 → 标题」顺序让步，至少保住 3 行选项。
 - assistant 正文在底部暂存窗比刷进 scrollback 后更亮：scrollback 正文自 2026-07-07 起不上前景色、
   跟随终端默认色，而暂存窗仍钦定近白 `#F1F4F8`，两边分家；且行内代码之后的字被 Rich 的
   恢复默认色指令拉回终端默认，同一行两档亮度。现在暂存窗正文同样交给终端默认前景色

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/CONTEXT.md

@@ -178,6 +178,20 @@ _Avoid_: window title, OSC title, process title (each names one channel, not the
 What the session currently needs from the user, as one of three values: idle (nothing running), busy (a turn or long command is running), needs-you (execution is paused on a tool approval or a question to the user). It is the only input that selects the Tab Title glyph.
 _Avoid_: spinner state, running flag, blocked
 
+## Model Tier Language
+
+**Provider-Pool Level**:
+The SDK's machine contract for the three model slots a session carries: `LOW`, `MID`, `HIGH`. It is what `settings.json` (`llm_levels*`), session metadata, the token ledger, subagent frontmatter and the JSON-RPC surface store and exchange. Never shown to the user by the CLI.
+_Avoid_: model level (user-facing), LLM level (user-facing label)
+
+**Model Tier**:
+The user-facing name of a Provider-Pool Level: **Lite** (`LOW`), **Pro** (`MID`), **Elite** (`HIGH`). A pure display vocabulary owned by the client, derived from one mapping module, never persisted and never sent to the SDK.
+_Avoid_: level (in any user-visible text), Max (OpenWorker's transitional name for `HIGH`, being retired), Ultra
+
+**Tier Tagline**:
+The one-line purpose statement shown next to a Model Tier wherever a model name follows it (tier menu, first-run wizard): Lite "Quick answers, small edits", Pro "Everyday coding", Elite "Deep reasoning, hardest problems". `/config` appends a cost clause to the same tagline. Markers such as "(default)" and "(current)" are appended by the display layer, never written into the tagline.
+_Avoid_: level description, help text
+
 ## Relationships
 
 - A **Core Slash Command** is always registered with `source="builtin"` on the Execution Axis. Its name belongs to the **Reserved Name Set** and cannot be overridden by any markdown-template command.
@@ -219,6 +233,11 @@ _Avoid_: spinner state, running flag, blocked
 - **Attention State** follows the TUI's single busy switch plus the two pause points (tool approval, question to the user). Background job completion and the exit confirmation dialog do not change it.
 - **Attention State** never animates: each state maps to exactly one glyph, so the terminal is written only on state changes and title changes.
 
+- A **Model Tier** is a rendering of a **Provider-Pool Level**, not a second identity: `/model lite` and `/model LOW` name the same slot, and the CLI translates user input back to the Provider-Pool Level before calling the SDK.
+- The word "tier" is reserved for user-visible text; the word "level" stays on the SDK contract, settings keys and RPC fields. Effort has its own ladder (`low`…`max`); a Model Tier name must never coincide with an effort level name.
+- **Tier Taglines** exist in exactly one place; every surface (tier menu, statusline, `/config`, wizard, preflight summary, endpoint warnings) derives from it.
+
 ## Flagged ambiguities
 
-- None currently. The original "builtin" collision between Execution Axis and Source Axis has been resolved by reserving "builtin" for Core Slash Commands and introducing "default" for the new shipped markdown resources.
+- OpenWorker still shows `HIGH` as **Max** (ADR-0017/0019); the agreed family-wide name is **Elite** (ADR-0023). Until OpenWorker is updated, the same Provider-Pool Level has two display names across products.
+- The original "builtin" collision between Execution Axis and Source Axis has been resolved by reserving "builtin" for Core Slash Commands and introducing "default" for the new shipped markdown resources.

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: comate-cli
-Version: 0.12.3
+Version: 0.12.4b1
 Summary: Comate terminal CLI built on comate-agent-sdk
 Project-URL: Homepage, https://github.com/AndyLee1024/agent-sdk
 Project-URL: Repository, https://github.com/AndyLee1024/agent-sdk
@@ -15,7 +15,7 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Requires-Python: <3.14,>=3.11
 Requires-Dist: charset-normalizer==3.4.7
-Requires-Dist: comate-agent-sdk<0.13,>=0.12.8
+Requires-Dist: comate-agent-sdk<0.13,>=0.12.9
 Requires-Dist: concurrent-log-handler>=0.9.25
 Requires-Dist: curl-cffi==0.13.0
 Requires-Dist: httpx<0.29.0,>=0.28.1

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/app.py

@@ -694,6 +694,7 @@ def _build_agent(
     system_role: str | None = None,
     yolo: bool = False,
     approval_bridge: Any | None = None,
+    prewarm: str = "thread",
 ) -> Agent:
     Agent, AgentConfig, _, EnvOptions = _import_sdk_agent_components(profiler)
     from comate_cli.terminal_agent.config import store
@@ -728,6 +729,7 @@ def _build_agent(
         # bypassPermissions for unattended embeddings, while CLI starts safe.
         permission_mode="bypassPermissions" if yolo else "default",
         tool_approval_callback=approval_bridge,
+        prewarm=prewarm,  # type: ignore[arg-type]
     )
     if profiler is not None:
         profiler.mark("agent.construct.start")
@@ -1253,6 +1255,16 @@ async def run(
         if interactive
         else nullcontext()
     )
+    # 预热档位按宿主形态选（rules/async-event-loop-blocking.md，实测 2026-09-21）：
+    # -p 没有「用户打字前的空闲窗口」，起线程只会与主线程抢 GIL → inline 同步预热；
+    # 交互 TUI 在 build_template 之后还有 TUI import + session 构造要在主线程跑，
+    # 线程起早了会把提示符从 1.08 s 拖到 2.35 s → deferred，首帧渲染后由 TUI 触发。
+    if print_message is not None:
+        prewarm_mode = "inline"
+    elif interactive:
+        prewarm_mode = "deferred"
+    else:
+        prewarm_mode = "thread"
     with agent_spinner:
         agent = _build_agent(
             project_root=project_root,
@@ -1264,6 +1276,7 @@ async def run(
             system_role=profile.system_role,
             yolo=yolo,
             approval_bridge=approval_bridge,
+            prewarm=prewarm_mode,
         )
     backend_runtime = None
     if profile.services is not None:
@@ -1389,6 +1402,8 @@ async def run(
         auto_title_controller=auto_title_controller,
         exit_epilogue=_exit_epilogue,
         profile=profile,
+        # deferred 档的预热入口；agent 缺少 prewarm（非 SDK 模板的替身）时 TUI 视为无预热。
+        prewarm=getattr(agent, "prewarm", None),
     )
     profiler.mark("tui.init.done")
     tui.add_resume_history(mode)

comate_cli-0.12.4b1/comate_cli/terminal_agent/bash_command_display.py

@@ -0,0 +1,79 @@
+"""Bash 命令的折叠显示：把多行脚本压成「前 N 条有效逻辑行」的单行片段。
+
+tool_view（scrollback / 进行中面板摘要）与 tool_fold（折叠 hint）共用这一份规则，
+两边都是纯展示，不参与权限分类——分类侧仍拿原始命令。
+
+「壳行」定义（只看行首尾字符，不做 shell 语法解析）：
+- 开壳行：引号 / 括号未闭合、以 ``\\`` 续行、或 heredoc 起始（``<<EOF``）。
+  它自己没有信息量，和下一行拼成同一条逻辑行，例如
+  ``python3 -c "`` + ``import json`` → ``python3 -c "import json``。
+- 闭壳行：整行只有收尾符号（``"``、``)``、``}`` 等）或 heredoc 结束标记，直接丢掉。
+"""
+from __future__ import annotations
+
+import re
+
+# heredoc 起始可以带后续重定向（`cat <<EOF > out.txt`），标记不一定在行尾。
+_HEREDOC_OPEN_RE = re.compile(r"<<-?\s*['\"]?(\w+)['\"]?")
+# 整行只有收尾符号（含分号）——这种行单独展示没有任何意义。
+_CLOSER_ONLY_RE = re.compile(r"""^[)\]}"';]+$""")
+_BRACKET_PAIRS = (("(", ")"), ("{", "}"), ("[", "]"))
+
+
+def _is_opener(line: str) -> bool:
+    if line.endswith("\\"):
+        return True
+    if _HEREDOC_OPEN_RE.search(line):
+        return True
+    for quote in ('"', "'"):
+        if line.endswith(quote) and line.count(quote) % 2 == 1:
+            return True
+    for open_ch, close_ch in _BRACKET_PAIRS:
+        if line.endswith(open_ch) and line.count(open_ch) > line.count(close_ch):
+            return True
+    return False
+
+
+def _merge(pending: str, line: str) -> str:
+    if pending.endswith("\\"):
+        return f"{pending[:-1].rstrip()} {line}"
+    if _HEREDOC_OPEN_RE.search(pending):
+        return f"{pending} {line}"
+    # 引号 / 括号开壳：直接贴上，保留 `python3 -c "import json` 的原貌
+    return f"{pending}{line}"
+
+
+def bash_logical_lines(content: str) -> list[str]:
+    """把命令拆成逻辑行：开壳行并入下一行，闭壳行丢弃。"""
+    logical: list[str] = []
+    pending = ""
+    heredoc_marker = ""
+    for raw in content.split("\n"):
+        line = raw.strip()
+        if not line:
+            continue
+        if heredoc_marker and line == heredoc_marker:
+            heredoc_marker = ""
+            continue
+        if _CLOSER_ONLY_RE.match(line):
+            continue
+        merged = _merge(pending, line) if pending else line
+        heredoc = _HEREDOC_OPEN_RE.search(line)
+        if heredoc:
+            heredoc_marker = heredoc.group(1)
+        if _is_opener(line):
+            pending = merged
+            continue
+        logical.append(merged)
+        pending = ""
+    if pending:
+        logical.append(pending)
+    return logical
+
+
+def collapse_bash_command(content: str, *, max_lines: int) -> tuple[str, bool]:
+    """取前 *max_lines* 条逻辑行拼成单行；第二个返回值表示有逻辑行被丢弃。"""
+    lines = bash_logical_lines(content)
+    if not lines:
+        return content.strip(), False
+    return " ".join(lines[:max_lines]), len(lines) > max_lines

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/config/picker.py

@@ -12,11 +12,8 @@ import logging
 from prompt_toolkit.layout.containers import HSplit, Window
 from prompt_toolkit.layout.controls import FormattedTextControl
 
-from comate_cli.terminal_agent.figures import (
-    EFFORT_HIGH,
-    EFFORT_LOW,
-    PLAY_ICON,
-)
+from comate_cli.terminal_agent.model_tier_style import tier_name
+from comate_cli.terminal_agent.figures import POINTER
 
 from .model import ConfigSnapshot
 from .picker_state import (
@@ -34,7 +31,7 @@ logger = logging.getLogger(__name__)
 _FIELD_LABELS: dict[ConfigField, str] = {
     ConfigField.DISPLAY_THINKING: "Thinking display",
     ConfigField.STREAM_PACING: "Reply pacing",
-    ConfigField.MODEL_LEVEL: "Default model level",
+    ConfigField.MODEL_LEVEL: "Default model tier",
     ConfigField.MEMORY_BACKGROUND: "Memory background",
     ConfigField.MEMORY_DREAMING: "Memory dreaming",
     ConfigField.TOKEN_COST: "Token cost",
@@ -128,9 +125,8 @@ class ConfigPickerUI:
         rows: list[tuple[str, str]] = []
         for f in _FIELD_ORDER:
             focused = f == state.focused_field
-            cursor = PLAY_ICON if focused else " "
-            marker = EFFORT_HIGH if focused else EFFORT_LOW
-            prefix = f"  {cursor} {marker} "
+            cursor = POINTER if focused else " "
+            prefix = f"  {cursor} "
             name = _FIELD_LABELS[f]
             value = self._render_field_value(f, state.draft)
             suffix = "  \u21b5" if f in _ENUM_FIELDS else ""
@@ -152,7 +148,7 @@ class ConfigPickerUI:
         if f == ConfigField.DISPLAY_THINKING:
             return f"[ {'on ' if snap.display_thinking else 'off'} ]"
         if f == ConfigField.MODEL_LEVEL:
-            return f"[ {snap.model_level} ]"
+            return f"[ {tier_name(snap.model_level)} ]"
         if f == ConfigField.STREAM_PACING:
             return f"[ {snap.stream_pacing} ]"
         if f == ConfigField.MEMORY_BACKGROUND:
@@ -176,9 +172,8 @@ class ConfigPickerUI:
         ]
         for idx, opt in enumerate(state.subview_options()):
             focused = idx == state.subview_cursor
-            cursor = PLAY_ICON if focused else " "
-            marker = EFFORT_HIGH if focused else EFFORT_LOW
-            prefix = f"  {cursor} {marker} "
+            cursor = POINTER if focused else " "
+            prefix = f"  {cursor} "
             style = (
                 "class:selection.option.selected"
                 if focused

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/config/picker_state.py

@@ -8,6 +8,7 @@ from dataclasses import dataclass, field
 from enum import Enum
 from typing import Callable
 
+from comate_cli.terminal_agent.model_tier_style import DEFAULT_LEVEL, tier_config_sentence, tier_name
 from comate_cli.terminal_agent.stream_pacer import STREAM_PACING_INTERVALS
 
 from .model import (
@@ -74,11 +75,11 @@ _STREAM_PACING_DESCRIPTIONS: dict[str, str] = {
     "normal": _ms_per_line("normal"),
     "fast": _ms_per_line("fast"),
 }
-_MODEL_LEVEL_DESCRIPTIONS: dict[str, str] = {
-    "LOW": "Fastest, cheapest. Good for simple tasks.",
-    "MID": "Default. Balanced speed and capability.",
-    "HIGH": "Most capable. For complex reasoning tasks.",
-}
+
+
+def _model_level_description(level: str) -> str:
+    sentence = tier_config_sentence(level)
+    return f"{sentence} (default)" if level == DEFAULT_LEVEL else sentence
 
 
 @dataclass
@@ -109,7 +110,7 @@ class ConfigPickerState:
             return []
         if self.subview_field == ConfigField.MODEL_LEVEL:
             return [
-                SubviewOption(v, v, _MODEL_LEVEL_DESCRIPTIONS[v])
+                SubviewOption(v, tier_name(v), _model_level_description(v))
                 for v in _MODEL_LEVEL_VALUES
             ]
         if self.subview_field == ConfigField.STREAM_PACING:

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/effort_menu.py

@@ -10,7 +10,12 @@ from typing import Any, Callable
 
 from comate_agent_sdk.agent.events import EffortSwitchedEvent
 from comate_agent_sdk.llm.effort import EFFORT_LADDER, EffortResolution, fold_down
-from comate_cli.terminal_agent.effort_style import ULTRACODE, effort_glyph, effort_style_str
+from comate_cli.terminal_agent.effort_style import (
+    ULTRACODE,
+    ULTRACODE_LEVEL,
+    effort_glyph,
+    effort_style_str,
+)
 from comate_cli.terminal_agent.i18n import i18n, message
 from comate_cli.terminal_agent.output_style_menu import SCOPE_TAGS
 
@@ -25,7 +30,7 @@ _OPTION_TEXT = {
 }
 
 # 刻度尺上的 7 格（左→右）；auto 不占格，用 `a` 键。ultracode 不是档位（rules/effort.md），
-# 在 CLI 层拆成 xhigh + 会话开关，只在这里当作第 7 格展示。
+# 在 CLI 层拆成 ULTRACODE_LEVEL + 会话开关，只在这里当作第 7 格展示。
 RULER_VALUES: tuple[str, ...] = ("off", *EFFORT_LADDER, ULTRACODE)
 
 
@@ -130,18 +135,22 @@ class EffortRulerSpec:
 
 
 def _level_note(level: str, resolution: EffortResolution) -> str:
-    """某一格的附注：不可关 / 会降级 / 不支持；正常时空串。ultracode 按 xhigh 判。"""
+    """某一格的附注：不可关 / 会降级 / 不支持；正常时空串。
+
+    ultracode 按 ULTRACODE_LEVEL 判降级；但模型完全不支持 effort 时它不是"不支持"——
+    workflows 这一半照样生效，只有思考力度落空，措辞必须分开。
+    """
     if level == "off":
         return "" if resolution.can_disable_thinking else i18n(message.effort_option_cannot_disable)
-    probe = "xhigh" if level == ULTRACODE else level
+    probe = ULTRACODE_LEVEL if level == ULTRACODE else level
     if probe in resolution.available_levels:
         return ""
     target = fold_down(probe, resolution.available_levels)  # type: ignore[arg-type]
-    return (
-        i18n(message.effort_option_downgrade, applied=target)
-        if target
-        else i18n(message.effort_option_unsupported)
-    )
+    if target:
+        return i18n(message.effort_option_downgrade, applied=target)
+    if level == ULTRACODE:
+        return i18n(message.effort_option_ultracode_effort_unsupported)
+    return i18n(message.effort_option_unsupported)
 
 
 def build_effort_ruler(
@@ -273,15 +282,18 @@ def _result_head_and_tail(
     """结果主行 + 可选的第二行提示。落点标签由调用方挂在主行末尾。"""
     prev = _label(event.previous, ultracode=ultracode_before)
     asked = _label(event.new, ultracode=ultracode_after)
-    # "谁压着"这一判据比较的是**档位**：ultracode 开着时请求档就是 xhigh，
-    # 显示成 ultracode 只是措辞，不能让 "xhigh" != "ultracode" 触发 kept 分支。
+    # "谁压着"这一判据比较的是**档位**：ultracode 开着时请求档就是 max，
+    # 显示成 ultracode 只是措辞，不能让 "max" != "ultracode" 触发 kept 分支。
     actual_requested = _label(resolution.requested, ultracode=ultracode_after)
 
     if not resolution.supported:
-        return (
-            i18n(message.effort_switched, previous=prev, new=asked),
-            i18n(message.effort_unsupported_saved),
+        # 开着 ultracode 时开关已经真的打开了，只有思考力度这一半落空：两件事都要说。
+        tail = (
+            i18n(message.effort_ultracode_on_effort_unsupported, level=ULTRACODE_LEVEL)
+            if ultracode_after
+            else i18n(message.effort_unsupported_saved)
         )
+        return i18n(message.effort_switched, previous=prev, new=asked), tail
 
     # 请求档没落到用户要的那个（最典型：env 层压着，切 auto 切不掉）
     if actual_requested != asked:

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/effort_ruler.py

@@ -361,7 +361,10 @@ class EffortRulerUI:
         """ultracode 选中时的彩色水波（背景色，rules/tui.md 具名例外）。"""
         src_col, front, spreading = ctx
         dist = math.hypot(col - src_col, row - _MENU_ROW)
-        if dist > front:
+        # front 恒为整数（帧 × _RIPPLE_SPEED），波源行的 dist 也恒为整数：若用 `>`，
+        # 波源行 dist == front 的那一格会比上下行（dist 带斜边、略大）多伸出 1 格，
+        # 白色波前在那一行变成 2 格，看起来像一个突出的方块。用 `>=` 把它拦掉。
+        if dist >= front:
             return None
         if spreading and front - dist <= 1.5:
             return theme.EFFORT_RULER_RIPPLE_FRONT

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/effort_style.py

@@ -20,8 +20,10 @@ from comate_cli.terminal_agent.figures import (
     EFFORT_OFF,
 )
 
-# 显示用伪档位：ultracode = xhigh + 会话开关（rules/effort.md），不进 EFFORT_LADDER。
+# 显示用伪档位：ultracode = max + 会话开关（rules/effort.md），不进 EFFORT_LADDER。
 ULTRACODE = "ultracode"
+# ultracode 绑定的真实档位；CLI 拆解 / 附注 / 关开关判定都只认这一个常量。
+ULTRACODE_LEVEL = "max"
 
 EFFORT_GLYPHS: dict[str, str] = {
     "off": EFFORT_OFF,

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/history_printer.py

@@ -25,6 +25,7 @@ from comate_cli.terminal_agent.message_style import (
     USER_TEXT_STYLE,
     tool_result_glyph_style,
 )
+from comate_cli.terminal_agent.ultracode_keyword import ultracode_keyword_spans
 from comate_cli.terminal_agent.tool_result_state import ToolResultState
 from comate_cli.terminal_agent.markdown_render import ANNOUNCEMENT_ACCENT
 from comate_cli.terminal_agent.jobs_state import job_display_label, job_event_detail
@@ -266,12 +267,25 @@ def _render_user_visual_line(
     # pad_width 非 None 时铺 CC 风格的满行底块（普通用户输入）；slash 命令行传 None。
     line_text = Text()
     line_text.append(prefix, style=USER_PREFIX_STYLE)
+    # 两类染色区间：skill slash（薰衣草紫）与裸词 ultracode（深紫，静态不流光——
+    # scrollback 是死文本）。按起点排序、重叠时先到者赢。
+    spans: list[tuple[int, int, str]] = [
+        (match.start(), match.end(), theme.SKILL_SLASH_STYLE)
+        for match in _SLASH_COMMAND_TOKEN_RE.finditer(visual_line)
+    ]
+    spans.extend(
+        (start, end, theme.ULTRACODE_KEYWORD_STYLE)
+        for start, end in ultracode_keyword_spans(visual_line)
+    )
+    spans.sort(key=lambda span: (span[0], -span[1]))
     cursor = 0
-    for match in _SLASH_COMMAND_TOKEN_RE.finditer(visual_line):
-        if match.start() > cursor:
-            line_text.append(visual_line[cursor : match.start()], style=USER_TEXT_STYLE)
-        line_text.append(match.group(0), style=theme.SKILL_SLASH_STYLE)
-        cursor = match.end()
+    for start, end, style in spans:
+        if start < cursor:
+            continue
+        if start > cursor:
+            line_text.append(visual_line[cursor:start], style=USER_TEXT_STYLE)
+        line_text.append(visual_line[start:end], style=style)
+        cursor = end
     if cursor < len(visual_line):
         line_text.append(visual_line[cursor:], style=USER_TEXT_STYLE)
     if pad_width is not None:

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/i18n/_en.py

@@ -61,7 +61,9 @@ EN: dict[str, str] = {
     "effort_value_auto": "auto",
     "effort_value_off": "off",
     "effort_value_ultracode": "ultracode",
-    "effort_option_ultracode": "xhigh + workflows: orchestrate every substantive task",
+    "effort_option_ultracode": "max + workflows: orchestrate every substantive task",
+    "effort_option_ultracode_effort_unsupported": "effort has no effect here; workflows only",
+    "effort_ultracode_on_effort_unsupported": "ultracode is on (multi-agent orchestration). The current model does not support effort; {level} is saved but will not be sent.",
     "effort_ruler_faster": "Faster",
     "effort_ruler_smarter": "Smarter",
     "effort_ruler_hint": "←/→ adjust · Enter confirm · s this session only · a auto · Esc cancel",

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/i18n/_ja.py

@@ -61,7 +61,9 @@ JA: dict[str, str] = {
     "effort_value_auto": "auto",
     "effort_value_off": "off",
     "effort_value_ultracode": "ultracode",
-    "effort_option_ultracode": "xhigh + workflows：実質的なタスクは既定でマルチエージェント編成",
+    "effort_option_ultracode": "max + workflows：実質的なタスクは既定でマルチエージェント編成",
+    "effort_option_ultracode_effort_unsupported": "思考量は効きません。workflows のみ",
+    "effort_ultracode_on_effort_unsupported": "ultracode を有効にしました（マルチエージェント編成）。現在のモデルは思考量に未対応のため、{level} は保存されますが送信されません。",
     "effort_ruler_faster": "速い",
     "effort_ruler_smarter": "賢い",
     "effort_ruler_hint": "←/→ 調整 · Enter 確定 · s このセッションのみ · a 自動 · Esc キャンセル",

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/i18n/_zh.py

@@ -52,7 +52,9 @@ ZH: dict[str, str] = {
     "effort_value_auto": "auto",
     "effort_value_off": "off",
     "effort_value_ultracode": "ultracode",
-    "effort_option_ultracode": "xhigh + workflows：实质任务默认多 agent 编排",
+    "effort_option_ultracode": "max + workflows：实质任务默认多 agent 编排",
+    "effort_option_ultracode_effort_unsupported": "思考力度不生效，仅 workflows",
+    "effort_ultracode_on_effort_unsupported": "ultracode 已开启（多 agent 编排）。当前模型不支持思考力度，{level} 已保存但不会发送。",
     "effort_ruler_faster": "更快",
     "effort_ruler_smarter": "更聪明",
     "effort_ruler_hint": "←/→ 调整 · Enter 确认 · s 仅本会话 · a 自动 · Esc 取消",

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/i18n/message.py

@@ -64,6 +64,8 @@ class message(StrEnum):  # noqa: N801 - 调用形态是 i18n(message.key)，小
     effort_value_off = "effort_value_off"
     effort_value_ultracode = "effort_value_ultracode"
     effort_ultracode_no_workflow_tool = "effort_ultracode_no_workflow_tool"
+    effort_ultracode_on_effort_unsupported = "effort_ultracode_on_effort_unsupported"
+    effort_option_ultracode_effort_unsupported = "effort_option_ultracode_effort_unsupported"
     effort_global_requires_level = "effort_global_requires_level"
     exit_requested = "exit_requested"
     feedback_description_prompt = "feedback_description_prompt"

{comate_cli-0.12.3 → comate_cli-0.12.4b1}/comate_cli/terminal_agent/jobs_picker.py

@@ -16,6 +16,7 @@ from prompt_toolkit.layout.controls import FormattedTextControl
 from prompt_toolkit.styles import Style
 
 from comate_cli.terminal_agent import theme
+from comate_cli.terminal_agent.figures import POINTER
 from comate_cli.terminal_agent.i18n import i18n, message
 from comate_cli.terminal_agent.jobs_state import (
     display_width,
@@ -216,7 +217,7 @@ def format_jobs_snapshot(items: Iterable[JobsPickerItem]) -> str:
 
 
 # 双通道配色（2026-09-03 picker 去青决策，与 workflows_panel 同处方）：
-# 状态色只落在状态符号与状态词上，选中只落在 › 与名字（白亮粗），
+# 状态色只落在状态符号与状态词上，选中只落在 ❯ 与名字（白亮粗），
 # 指标尾巴恒 faint——禁止整行大色块。
 _STYLE = Style.from_dict(
     {
@@ -360,7 +361,7 @@ class JobsPickerApp:
         fragments: list[tuple[str, str]] = []
         for index, item in enumerate(self.items):
             selected = index == self._selected_index
-            marker = "›" if selected else " "
+            marker = POINTER if selected else " "
             symbol, status_style = _job_visual(item)
             status = "stopping…" if item.stopping else item.status
             if selected: