@aragon-agent/cli 0.6.12 → 0.6.14

package/CHANGELOG.md

@@ -7,6 +7,33 @@ were written.
 
 ## Unreleased
 
+### Consistent TUI styling
+
+- Interactive sessions always use Unicode and default to truecolor, including
+  PowerShell sessions without terminal or UTF locale markers. Ink's actual color
+  output follows the same policy, with renderer state restored on exit.
+- Explicit color opt-outs keep Unicode borders and icons; the opening logo now
+  depends only on available space. Headless terminal detection is unchanged.
+- Reloading model settings preserves session display preferences. Syntax
+  highlighting follows the same color policy, and status icons use consistent
+  single-column glyphs to prevent narrow-frame overflow.
+- Hiding the editor caret while scrolling preserves draft geometry, preventing
+  a layout update loop when entering text above the full startup logo.
+- Register exit cleanup before mounting the application, so immediate exits and
+  initial render failures restore terminal and color state reliably. Failed
+  mounts no longer mark an auto-installed build as healthy.
+
+### Model context window
+
+- Discover context limits at startup and after connection changes without blocking
+  input. The status bar and compaction use the discovered limit; `/context` names
+  API versus catalog metadata. Unknown windows show `?`, not a fabricated 128k
+  capacity or percentage. Manual `contextWindow` overrides retain precedence.
+- Window overrides now apply to tail relief and compaction progress as well as
+  the gauge. Workers on different models/endpoints do not inherit the lead's
+  override. API-key changes refresh metadata, and ending a session cancels its
+  pending model queries. `/context` also hides unknown capacities/percentages.
+
 ### TUI 输入与排队可靠性
 
 - Shift+Enter 与粘贴按输入顺序处理；普通 Enter 保持弹层兼容，混合输入最多提交一次，

package/README.md

@@ -432,8 +432,8 @@ completes, `Up` / `Down` moves the selection, `Esc` closes it.
 
 | Key | Action |
 | --- | --- |
-| `Enter` | Submit (idle) / queue a steering message (running) — shown as `Queue: …` until the agent picks it up |
-| `Ctrl+J` / `Alt+Enter` / `Shift+Enter` | Insert a newline. `Ctrl+J` and `Alt+Enter` work everywhere; a distinct `Shift+Enter` needs a one-line terminal binding — see [`/terminal-setup`](#terminal-setup) |
+| `Enter` | 空闲时发送，运行时加入队列；收到 Core 已纳入会话的接收凭据后才移除 Queue 提示 |
+| `Ctrl+J` / `Alt+Enter` / `Shift+Enter` | 插入换行。Ctrl+J 已通过输入链路测试；另外两种组合键需要宿主发送独立序列，Alt+Enter 可能被宿主拦截。参见 [终端设置](#terminal-setup-for-shiftenter) |
 | `Shift+Tab` | Toggle **plan mode** (`BUILD` <-> `PLAN`). Equivalent: `/plan`. On Windows this needs a console flag Node < 22.17 does not set; aragon sets it at startup — see [Requirements](#requirements) |
 | `Ctrl+P` | The same toggle, over a channel no Windows console can swallow. Use it if `Shift+Tab` does nothing |
 | `Esc` | Close an overlay or popup; otherwise ask for interruption confirmation. While idle, cancel pending auto-continuation |
@@ -452,7 +452,7 @@ completes, `Up` / `Down` moves the selection, `Esc` closes it.
 | `Alt/Ctrl+←` / `→` | Word-wise cursor jump |
 | `Ctrl+W`, `Alt+Backspace` | Delete the previous word |
 | `Ctrl+U` / `Ctrl+K` | Kill to line start / end |
-| `Up` / `Down` | Move between draft lines; recall prompt history at the edges |
+| `Up` / `Down` | 按视觉行移动；空草稿或正在浏览历史时，才在边界召回历史，避免覆盖新草稿 |
 | Paste | Inserted as-is up to 6 lines / 400 chars; larger collapses to `[Pasted text #1 +15 lines]`, which deletes as one unit and expands in full when you send. `--no-paste` restores the pre-0.6.3 behaviour |
 | `?` | Open help (empty input) |
 
@@ -462,7 +462,7 @@ new output automatically; once you scroll away the status bar shows `↑N`, a sl
 scrollbar occupies the terminal's last column (including 40-column windows). Drag its thumb
 to move through history, or click the track to move one page. Messages, the team panel
 and the input area scroll together; typing or editing returns to the input without
-losing the draft. The fixed status row shows `PgDn down` while reading history.
+losing the draft. The fixed status row reports the distance from the end while reading history.
 `PgDn` moves one page down; editing or submitting returns directly to the end.
 
 **The mouse wheel is a viewport gesture and nothing else.** Wherever the pointer
@@ -498,10 +498,9 @@ Shift+Enter 与 Enter 都发送为 CR，应用无法区分；请用 `/terminal-s
 让 Shift+Enter 发送 `\u001b[13;2u`。合并到已有终端配置，不要覆盖其他快捷键；
 终端拦截某个组合键时可使用另外两种方式。参考绑定如下：
 
-- **Windows Terminal** — `settings.json`, `actions`:
-  `{ "keys": "shift+enter", "command": { "action": "sendInput", "input": "\u001b[13;2u" } }`
+- **Windows Terminal** — `settings.json` 的 `actions` 与 `keybindings` 分别定义动作和按键；完整示例见下文“输入、待处理队列与底部状态区”。
 - **VS Code integrated terminal** — `keybindings.json`:
-  `{ "key": "shift+enter", "command": "workbench.action.terminal.sendSequence", "args": { "text": "\u001b[13;2u" } }`
+  `{ "key": "shift+enter", "command": "workbench.action.terminal.sendSequence", "when": "terminalFocus", "args": { "text": "\u001b[13;2u" } }`
 - **iTerm2** — Settings > Profiles > Keys > Key Mappings, `Shift+Enter` ->
   *Send Escape Sequence* `[13;2u`
 - **kitty / WezTerm / foot** — native once keyboard-enhancement mode is on
@@ -518,9 +517,9 @@ user-side binding reaches the same result for the one key that needs it.
 
 ## Queued messages while the agent runs
 
-Agent 运行中发送的消息会进入队列。固定状态行显示最早待接收消息的首个非空行，
-多条时显示剩余数量；即使转录滚动离开底部，提示仍可见，且没有自动消失的时间限制。
-转录同时保留每条完整消息：
+Agent 运行中发送的消息会进入队列。输入框附近的独立 Queue 面板按顺序显示摘要，
+限高后显示隐藏条数；固定状态栏始终显示真实待处理总数，没有自动消失的时间限制。
+使用 `/queue` 分页查看所有消息的完整原文，未接收消息不会在普通转录中重复显示：
 
 ```
 ◷ Queue: also check the tests
@@ -528,11 +527,9 @@ Agent 运行中发送的消息会进入队列。固定状态行显示最早待
 
 仅当 Core 将消息写入对话历史并发出对应消息 ID 的接收回执，Queue 状态才移除。
 这表示引擎已接管后续处理，不表示回答完成；`turn_start`、动画和提示超时都不能代替回执。
-相同正文的两次发送仍是两条消息。中断或失败后尚未接收的消息显示 `Queue (paused)`。
-
-窄屏使用 `P>B` / `B>P` 表示待切换模式，`E2` / `E!` 表示两次 Esc 中断 / 再按 Esc 强停，
-`S+` / `T+` 表示后台服务 / TODO 存在，`Queue(p)` 表示暂停；`(+99+)` 表示至少还有
-100 条消息。完整服务和 TODO 数量分别可用 `/bg` 与 `/todo status` 查看。
+相同正文的两次发送仍是两条消息。中断或失败后尚未接收的消息保留并标记暂停。
+窄屏优先保留运行状态、待处理数量和上下文；其他字段整项隐藏，不截断关键数值。
+快捷键集中在操作提示行，并随覆盖层、补全、选区和中断确认状态变化。
 
 `/clear` 清转录和 TODO，但保留待接收队列。保存和退出会补回被清屏或保留环裁剪的
 未发送全文。恢复存档时，queued 一律成为 `Queued but never sent: ...` 警告，
@@ -1319,9 +1316,9 @@ The readouts drop as the terminal narrows, in that order of priority:
   either no completed turn has reported usage yet (right after `/resume`, or
   right after a compaction), or tool results have been appended since the last
   measurement and their size is estimated.
-- `?` sits on the **window** and means the *denominator* is a guess: the model is
-  not in the built-in table, so a flat 128 000 was substituted. See
-  `contextWindow` below.
+- `?` replaces the **window** when neither the API nor the catalog provides a
+  limit. No percentage is shown in that case. The internal 128 000-token fallback
+  remains a compaction budget, not a claim about model capacity. See `contextWindow` below.
 
 **The percentage keeps moving during a turn.** Every tool result nudges it,
 whether or not compaction is enabled — it is not a compaction feature, and a
@@ -1346,7 +1343,7 @@ Context
 ```
 
 Two lines are worth reading closely. **`Window`** always names where the number
-came from — the model table, your own `contextWindow`, or the invented
+came from — the model API, the model table, your own `contextWindow`, or the invented
 placeholder — because those are indistinguishable everywhere else. And
 **`Compaction`** reports what is actually true of this session, including
 `off - not registered for this session (started with --no-compaction)`; it never
@@ -1354,13 +1351,21 @@ promises a threshold that nothing will act on.
 
 ### `contextWindow`
 
-When the model is not in the built-in table — a custom `baseUrl`, a self-hosted
-id, something released last week — the denominator is a **fabricated** 128 000
-and the percentage is meaningless. Correct it:
+Window resolution uses your explicit override first, then API metadata from the
+configured endpoint, then the built-in catalog (including known aliases and dated
+snapshots). Startup, connection changes, and subsequent prompts refresh discovery
+in the background, with a five-minute cache. OpenAI-compatible endpoints may expose
+`context_length` or `context_window`; Anthropic uses `max_input_tokens`; Google uses
+`inputTokenLimit`. An ID-only models response cannot establish the real limit.
+Google/Anthropic lists are paginated (up to 20 pages within one 15-second deadline).
+Changing the connection or ending the session cancels its pending lookup.
+
+If neither the endpoint nor the catalog knows your model, the status bar shows `?`.
+Set a verified provider limit explicitly:
 
 ```bash
 aragon config set contextWindow 1000000   # persisted; prints "Set contextWindow = 1000000"
-aragon config set contextWindow auto      # back to the model table / placeholder
+aragon config set contextWindow auto      # back to API discovery / catalog
 ARAGON_CONTEXT_WINDOW=1000000 aragon      # one session
 ```
 
@@ -1893,7 +1898,7 @@ env / `.env` → CLI flags**.
 | Config key | Default | Meaning |
 | --- | --- | --- |
 | `maxTokens` | `64000` | Output token cap, clamped to `[256, 200000]`. `null` means **auto** — use each model's own ceiling, never above 64000. Absent means the default. See [Output token limits](#output-token-limits). |
-| `contextWindow` | `null` | The window the context gauge measures against, clamped to `[8000, 5000000]`. `null` means **auto** — the built-in model table, or a fabricated 128000 for a model it has never seen. Set it when the gauge shows `?` on the denominator. See [The context gauge](#the-context-gauge). |
+| `contextWindow` | `null` | The window the context gauge measures against, clamped to `[8000, 5000000]`. `null` means **auto** — API metadata, then the built-in catalog. Unknown windows show `?`; the internal fallback is 128000. See [The context gauge](#the-context-gauge). |
 | `showThinking` | `false` | Draw the reasoning the model returns. `thinkingLevel` is the effort the provider is asked to **spend**; this is whether the terminal **shows** it. Off by default: a settled turn that thought leaves one muted `thought for 12s` row in its place, so nothing is hidden silently. |
 | `liveToolOutput` | `true` | Draw up to eight sanitised rows of a **running** tool's output on its card, plus a `no output for Ns` row when the child goes quiet. On by default, unlike `showThinking`: this ADDS the information a long `bash` call otherwise hides, and its cost is bounded by construction — eight rows per call, sixteen calls, whatever the command emits. Turn it off and the card is a single `running` row again, with no store allocated and no recorder attached. |
 | `exitTranscript` | `true` | Replay a plain-text session summary after exiting (full-screen only). |
@@ -1971,7 +1976,7 @@ aragon config set maxTokens 32000    # persisted; prints "Set maxTokens = 32000"
 aragon config set maxTokens auto     # persisted as null
 aragon config set maxTokens default  # back to 64000
 aragon config set contextWindow 1000000  # the gauge's denominator
-aragon config set contextWindow auto     # back to the model table
+aragon config set contextWindow auto     # back to API discovery / catalog
 aragon config get maxTokens          # "64000" | "auto"
 ```
 
@@ -2275,12 +2280,15 @@ quiet output is required.
 `--theme` / `config.theme` / `/theme` pick a palette: `warm`, `cool`, `light`, or
 `auto` (which resolves to `warm` — terminals can't reliably report their
 background). `dark` remains accepted everywhere as an alias for `cool`.
-Colors degrade automatically to your terminal's depth (truecolor → 256 → 16 →
-monochrome) and glyphs fall back to ASCII on terminals without Unicode. Color is
-disabled — and the UI renders plain monochrome with ASCII glyphs — when any of
-`NO_COLOR`, `--no-color`, `TERM=dumb`, or `config.color=false` is set; `FORCE_COLOR`
-is respected. `reducedMotion` (also `ARAGON_REDUCED_MOTION`, and implied by
-`--no-color`) replaces spinners with a static glyph.
+Interactive sessions always use Unicode borders, icons and scrollbars, with
+truecolor and gradients by default. Missing terminal/locale markers, `TERM=dumb`,
+and low `FORCE_COLOR` levels do not simplify the TUI. `--no-color`, `NO_COLOR`
+(including an empty value), or `FORCE_COLOR=0/false` explicitly disable color;
+Unicode stays enabled. The opening logo adapts only to available space and
+remains visible in monochrome. `reducedMotion` (also `ARAGON_REDUCED_MOTION`,
+and implied by disabling color) replaces spinners with a static glyph.
+Headless commands retain their existing terminal detection. The terminal and
+font still need to support the characters and colors the TUI emits.
 
 ### Timeout invariant
 
@@ -2345,3 +2353,72 @@ UTF-16 编辑和 ZWJ/组合字形仍有既有限制，本次修复保证视觉
 检查流式输出不阻止闪烁；单 ESC 不停止任务，两击中断，卡住时第三击强停；菜单关闭不计入两击。
 任务结束后保留最终计划，下一条普通消息启动时旧计划消失，新有效计划出现才恢复；显式续跑保留计划。
 强停后立即提交，在等待中双 ESC 取消，旧引擎退出后不得自行执行已取消的消息。
+
+## 输入、待处理队列与底部状态区
+
+Enter 在空闲时发送，运行时加入队列；启动阶段拒绝重复提交并保留草稿。
+Ctrl+J 换行；终端能发送独立序列时，Shift+Enter 与 Alt+Enter 也可换行。
+光标按完整字素移动，支持中文、组合字符和 emoji；输入最多占终端高度四分之一且不超过六行。
+括号粘贴保留换行，不触发发送。过滤器或粘贴支持关闭时，不保证未标记多行粘贴安全。
+过滤器不可用时 Delete 保守按 Backspace 处理，避免退格误删后文。
+
+应用鼠标拖选只建立选区，Ctrl+C 才复制；复制中再次按 Ctrl+C 不会退出或停止服务。
+原生剪贴板工具完成后显示确认，OSC 52 只表示请求已发送，终端可能拒绝。
+Shift+拖动或 `/mouse off` 后的选区由宿主管理，应用无法关闭宿主的自动复制。
+
+Queue 面板逐条显示待处理消息，数量始终留在状态栏。只有 Core 已把消息纳入会话并发回接收凭据，
+该条提示才会消失。中断时未接收消息显示暂停；`/queue` 可分页查看全部原文，Esc 关闭。
+`/clear` 不丢待处理消息；`/reset`、切换会话会明确报告取消数量；恢复保存的未发送历史不会自动重发。
+
+底部依次为输入框、固定操作提示/短反馈行、全局状态栏。状态字段按显示宽度完整保留或整项隐藏。
+上下文表示当前主 Agent 的已用/上限 tokens；`~` 表示含估算，`?` 表示上限未知。
+费用是含子代理、快速模型及压缩的会话估算 USD，恢复会话沿用累计，缺价格按零计，零值不保证免费。
+本轮均速是本轮期间入账输出 token 增量除以耗时，可能包含上述辅助调用，并非瞬时生成速度。
+`/context` 提供统计范围和完整值。
+
+Windows Terminal：在设置 JSON **根级**添加 `"copyOnSelect": false`，把以下条目合并到已有数组：
+
+```json
+{
+  "actions": [{ "id": "User.AragonNewline", "command": {
+    "action": "sendInput", "input": "\u001b[13;2u"
+  } }],
+  "keybindings": [{ "keys": "shift+enter", "id": "User.AragonNewline" }]
+}
+```
+
+依据：[Windows Terminal 交互设置](https://learn.microsoft.com/en-us/windows/terminal/customize-settings/interaction)、
+[按键动作](https://learn.microsoft.com/en-us/windows/terminal/customize-settings/actions)。PowerShell 是 shell，不能在其 profile 中设置终端按键映射。
+
+VS Code 用户设置添加 `"terminal.integrated.copyOnSelection": false`，快捷键 JSON 添加：
+
+```json
+{
+  "key": "shift+enter", "command": "workbench.action.terminal.sendSequence",
+  "when": "terminalFocus", "args": { "text": "\u001b[13;2u" }
+}
+```
+
+依据：[终端基础](https://code.visualstudio.com/docs/terminal/basics)、
+[终端高级设置](https://code.visualstudio.com/docs/terminal/advanced)。应用的 `/terminal-setup` 只输出说明，不修改宿主配置。
+若宿主把 Shift+Enter 或 IME 候选确认同样编码为 CR，应用无法从相同字节推断用户意图。
+可使用经输入链路测试的 Ctrl+J、独立 CSI-u 绑定及外部编辑后括号粘贴；具体宿主组合仍需实测。
+双宿主、微软拼音及持续流式性能验收状态见 `docs/plans/tui-composer-queue-status/manual-test.md`。
+
+## 多套模型配置
+
+在 `/settings` 中使用 **Main profile** 和 **Fast profile**，可以分别为主任务与快速模型选择已保存的方案。两种角色共享配置库，但各自选择互不影响，也可以选择同一个方案。**Manage profiles** 提供新建、编辑、复制和删除；名称留空时采用模型名。同名方案通过地址主机和唯一 ID 区分。超过八条时按 `/` 搜索名称、模型或服务商。
+
+上下方向键或 Tab 移动，Enter 选择或编辑，根设置页 Ctrl+S 保存。普通文本字段的 Enter 仍保存设置。选择和子表单的修改先进入草稿；Esc 返回或确认丢弃。被主模型或快速模型引用的方案必须先解除绑定才能删除。复制包含本轮已经编辑的凭据，并产生新的 ID。
+
+方案保存服务商、模型、地址和凭据，不改变思考级别、token 上限、快速评审频率或预算。选择快速模型方案不会自动启用快速层。若本次启动时快速层未注册，首次启用会提示“已保存，待重启”；缺少模型或凭据时也需要先补齐。
+
+**Current custom** 保留旧式连接，可继续编辑原连接字段。创建方案时从磁盘上的旧式配置预填，不复制环境变量或 `--api-key` 的运行时密钥。专属密钥可保持、替换或切换为共享/环境凭据；修改服务商或地址后，必须明确选择保留原密钥，或重新输入。空地址在方案中表示服务商默认地址，快速方案不会继承主网关。
+
+启动优先级为 CLI 参数、环境变量、所选方案、旧式字段、默认值。界面显示启动覆盖；明确重新选择方案可对本次会话立即生效，重启仍遵循启动参数。专属密钥仅对对应角色和目标地址生效；主模型启动密钥不会跟随切换发往另一个网关。
+
+运行、团队任务、后台评审或压缩尚未结束时禁止保存。保存采用原子文件替换，同时检查磁盘与当前会话版本；失败保留草稿，发生冲突时选择 Reload settings。写入已成功但会话应用异常时会明确提示重启并禁用重复保存。跨进程版本检查属于乐观检查，不能消除同时写入的极小竞态。
+
+`/model`、`/fast model` 和 `config set` 修改连接会回到对应角色的自定义配置。若解除绑定会丢失专属账户或默认地址语义，命令会拒绝并引导到设置中编辑或复制。`/reload` 会一起刷新两个角色的连接和凭据，保留对话与用量。
+
+配置库仍位于 `ARAGON_HOME` 指向目录中的 `config.json`，不写入会话文件。密钥按现有机制明文保存：POSIX 创建权限为 0600，Windows 依赖用户目录 ACL，不提供加密。`config get/list` 对方案密钥使用固定掩码；损坏或未来版本的配置段整体隐藏。既有 `config edit` 备份也包含凭据。关闭通用日志脱敏、使用短密钥或分享第三方 SDK 错误日志时，仍需自行检查日志内容。

package/dist/agent/controller.d.ts

@@ -1,3 +1,5 @@
+import { type ModelSettingsSnapshot } from './model-profile-settings.js';
+import type { ModelSettingsDraft, ModelSettingsPatch, ModelSettingsSaveResult } from '../config/model-profile-store.js';
 /**
  * AgentController — builds the core `Agent`, wires the built-in toolset, and
  * exposes a small run/abort/steer surface plus the pre-flight validation that
@@ -11,6 +13,7 @@
 import type { PromptOptions, PromptOutcome } from './prompt-options.js';
 import { ModelRegistry, type AgentEvent, type AgentTool, type Message, type ModelInfo, type ModelRef, type ProviderRegistry, type ThinkingLevel } from '@aragon-agent/core';
 import type { CliConfig, CompactionConfig, FastConfig, RetryConfig, TeamConfig, ThemeName, TodoConfig } from '../config/schema.js';
+import type { ModelRole } from '../config/model-profiles.js';
 import { type ConfirmRequest } from '../tools/index.js';
 import type { ToolPermission } from '../exec/permission.js';
 import { type ToolOutputListener } from '../tools/tool-output-store.js';
@@ -120,6 +123,10 @@ export declare class AgentController {
     private readonly agent;
     private readonly providerRegistry;
     private readonly modelRegistry;
+    private modelMetadataEnabled;
+    private modelMetadataGeneration;
+    private modelMetadataAbort;
+    private disposed;
     private readonly tools;
     private readonly skills;
     private readonly skillsEnabled;
@@ -180,6 +187,9 @@ export declare class AgentController {
     private readonly outputListeners;
     private readonly fast;
     private readonly compaction;
+    private settingsRevision;
+    private modelSettingsBlocked;
+    private preparedSystemPrompt;
     private readonly contextMeter;
     private readonly procs;
     /**
@@ -249,7 +259,7 @@ export declare class AgentController {
      * Read-only CLI-layer helper for the Header/Welcome key-status dot (P1-1);
      * does not touch `@aragon-agent/core`.
      */
-    hasApiKey(provider?: string): boolean;
+    hasApiKey(provider?: string, role?: ModelRole): boolean;
     /**
      * Set the color theme live. Mirrors the `setModel` / `setThinkingLevel`
      * mutator pattern — a subsequent re-render re-reads `getConfig().theme` and
@@ -524,6 +534,28 @@ export declare class AgentController {
      * the runtime nor the file.
      */
     setRetryConfig(patch: Partial<RetryConfig>): RetryConfig;
+    /** True while any request can still use a model or credential snapshot. */
+    isModelSettingsBusy(): boolean;
+    /** Monotonic revision protects an editor from all intervening settings changes. */
+    getSettingsRevision(): number;
+    /** Read-only draft creation; no startup migration or credential copying. */
+    createModelSettingsDraft(): ModelSettingsDraft;
+    /** Save an explicit draft as a single disk/live transaction. */
+    saveModelSettings(draft: ModelSettingsDraft): ModelSettingsSaveResult;
+    /** Legacy commands share the same lossless conversion and atomic save path. */
+    saveModelSettingsPatch(patch: ModelSettingsPatch): ModelSettingsSaveResult;
+    /** Refresh both role connections and credentials while retaining conversation state. */
+    reloadModelSettings(): ModelSettingsSaveResult;
+    /** Prebuild all potentially failing model/prompt work before committing the file. */
+    prepareModelSettingsSnapshot(config: CliConfig): ModelSettingsSnapshot;
+    /** Apply prepared state synchronously, without rebuilding the controller or its history. */
+    applyModelSettingsSnapshot(snapshot: ModelSettingsSnapshot): void;
+    /** Potentially failing ordinary-setting side effects are isolated from model application. */
+    applyModelSettingsEffects(): void;
+    /** A partially applied model snapshot must never reach another provider request. */
+    blockModelSettingsRequests(): void;
+    /** The UI uses this before accepting or draining queued prompts. */
+    areModelSettingsBlocked(): boolean;
     setModel(provider: string, model: string, baseUrl?: string): void;
     setThinkingLevel(level: ThinkingLevel): void;
     setMaxTokens(value: number | undefined): void;
@@ -580,7 +612,11 @@ export declare class AgentController {
      * cost table for the model that actually billed" cannot drift from "the cost
      * table for the session's model" as two separate bodies.
      */
-    getModelInfoFor(ref: Pick<ModelRef, 'providerId' | 'modelId'>): ModelInfo;
+    getModelInfoFor(ref: Pick<ModelRef, 'providerId' | 'modelId' | 'baseUrl'>): ModelInfo;
+    /** Best-effort metadata lookup; never block typing or a model request. */
+    refreshModelMetadata(): Promise<void>;
+    /** Invalidate an old connection without restarting a lookup on every prompt. */
+    private resetModelMetadataRequest;
     /**
      * Whether the STATIC table knows this model's price (C-11 / RV-4).
      *