dominds 0.9.2 → 0.9.3

package/dist/docs/team-mgmt-toolset.md

@@ -1,6 +1,6 @@
-# Team Management Toolset (`team-mgmt`)
+# Team Management Toolset (`team_mgmt`)
 
-Chinese version: [中文版](./team-mgmt-toolset.zh.md)
+Chinese version: [中文版](./team_mgmt-toolset.zh.md)
 
 This document specifies a dedicated **team management toolset** whose only job is managing the
 rtws’s “mindset” configuration files under `.minds/` (team roster, LLM providers, and agent
@@ -29,7 +29,7 @@ equivalent of the `ws_mod` toolset + unrestricted `read_dirs`/`write_dirs`), bec
 
 ## Migration Plan (Replacing legacy builtin team-manager knowledge)
 
-This document is a **design spec** for the new `team-mgmt` toolset. It is not something we should
+This document is a **design spec** for the new `team_mgmt` toolset. It is not something we should
 ever tell an agent to “look up” at runtime.
 
 Instead, the runtime “single source of truth” for team management guidance should be
@@ -75,9 +75,9 @@ This makes sense for “normal” agents, but it blocks the team manager from do
 - Providing general-purpose file editing across the repo.
 - Making `.minds/` broadly writable by default team members.
 
-## Proposed `team-mgmt` Toolset
+## Proposed `team_mgmt` Toolset
 
-The `team-mgmt` toolset mirrors a minimal subset of `fs`/`txt`, but **hard-scopes** all operations to
+The `team_mgmt` toolset mirrors a minimal subset of `fs`/`txt`, but **hard-scopes** all operations to
 `.minds/` and rejects anything outside.
 
 ### Naming Conventions (Human / UI)
@@ -124,7 +124,7 @@ Notes:
   - Reject paths containing `..`.
   - Reject any path that resolves outside `.minds/` after normalization.
 - Prefer an explicit allowlist over “anything in the rtws”.
-  - For `team-mgmt`, that explicit allowlist is `.minds/**` (including `.minds/memory/**`) so the
+  - For `team_mgmt`, that explicit allowlist is `.minds/**` (including `.minds/memory/**`) so the
     team manager can repair accidental corruptions made by other tools (even though `.minds/memory/**`
     already has dedicated `memory` / `team_memory` tools for normal use).
 - Require explicit `.minds/...` paths and validate them; do not support “implicitly scoped” paths
@@ -133,7 +133,7 @@ Notes:
 ### Why a dedicated toolset (instead of only `read_dirs` / `write_dirs`)?
 
 `read_dirs` / `write_dirs` are still valuable, but they are configured in `.minds/team.yaml`, which
-may not exist during bootstrap. A dedicated `team-mgmt` toolset:
+may not exist during bootstrap. A dedicated `team_mgmt` toolset:
 
 - Lets the team manager create `.minds/team.yaml` safely from “zero state”.
 - Keeps the scope bounded even if the member’s directory allow/deny lists are empty.
@@ -387,7 +387,7 @@ member_defaults:
     - ws_read
     - memory
   # Default posture: deny `.minds/` edits for normal members.
-  # (Team management should be done via `team-mgmt` tools, not general file tools.)
+  # (Team management should be done via `team_mgmt` tools, not general file tools.)
   no_read_dirs:
     - .minds/team.yaml
     - .minds/llm.yaml
@@ -426,7 +426,7 @@ Best practices:
 - Keep `.minds/team.yaml` ownership tight; only the team manager should be able to edit it.
 - Avoid repeating built-in constraints in `team.yaml`:
   - `*.tsk/**` (encapsulated Taskdocs) are hard-denied for all general file tools.
-  - `.minds/**` is hard-denied for general file tools; only the dedicated `team-mgmt` toolset can access it.
+  - `.minds/**` is hard-denied for general file tools; only the dedicated `team_mgmt` toolset can access it.
     Put these in `no_*` only when you need extra explicitness; they are enforced regardless.
 
 ## Managing `.minds/team/<member>/*.md` (agent minds)
@@ -460,8 +460,8 @@ Suggested structure:
 
 Preferred behavior for initial bootstrap:
 
-- The shadow `fuxi` instance should get `team-mgmt` (and the manual tool), not broad `ws_mod`.
-- The shadow `pangu` instance should get broad rtws toolsets (e.g. `ws_read`, `ws_mod`, `os`), but not `team-mgmt`.
+- The shadow `fuxi` instance should get `team_mgmt` (and the manual tool), not broad `ws_mod`.
+- The shadow `pangu` instance should get broad rtws toolsets (e.g. `ws_read`, `ws_mod`, `os`), but not `team_mgmt`.
 - After `.minds/team.yaml` is created, the team definition becomes the source of truth.
 
 This avoids needing to grant full rtws access to configure the team.
@@ -472,7 +472,7 @@ This avoids needing to grant full rtws access to configure the team.
   `member_defaults.model`.
 - **Provider not found**: Ensure `.minds/team.yaml` `provider` keys exist in merged provider config
   (`dominds/main/llm/defaults.yaml` + `.minds/llm.yaml`).
-- **Access denied when editing `.minds/`**: Intended for general file tools; use `team-mgmt` tools.
+- **Access denied when editing `.minds/`**: Intended for general file tools; use `team_mgmt` tools.
 - **MCP tools not visible in Tools view**:
   - Confirm `.minds/mcp.yaml` exists and is valid.
   - Open **Problems** and look for MCP-related errors.

package/dist/docs/team-mgmt-toolset.zh.md

@@ -1,6 +1,6 @@
-# 团队管理工具集（`team-mgmt`）
+# 团队管理工具集（`team_mgmt`）
 
-英文版：[English](./team-mgmt-toolset.md)
+英文版：[English](./team_mgmt-toolset.md)
 
 本文档指定了一个专用的**团队管理工具集**，其唯一职责是管理 rtws（运行时工作区）`.minds/` 下的"心智"配置文件（团队名单、LLM 提供商和智能体心智文件），而不授予广泛的运行时工作区访问权限。
 
@@ -23,7 +23,7 @@
 
 ## 迁移计划（替换传统的内置团队管理者知识）
 
-本文档是新的 `team-mgmt` 工具集的**设计规范**。这不是我们应该在运行时让智能体"查阅"的内容。
+本文档是新的 `team_mgmt` 工具集的**设计规范**。这不是我们应该在运行时让智能体"查阅"的内容。
 
 相反，运行时团队管理的"单一事实来源"应该是函数工具 `team_mgmt_manual` 的输出。
 
@@ -61,9 +61,9 @@
 - 提供跨仓库的通用文件编辑
 - 默认让 `.minds/` 可广泛写入
 
-## 提议的 `team-mgmt` 工具集
+## 提议的 `team_mgmt` 工具集
 
-`team-mgmt` 工具集镜像 `fs`/`txt` 的最小子集，但**硬作用域**所有操作到 `.minds/` 并拒绝任何外部操作。
+`team_mgmt` 工具集镜像 `fs`/`txt` 的最小子集，但**硬作用域**所有操作到 `.minds/` 并拒绝任何外部操作。
 
 ### 命名约定（人类 / UI）
 
@@ -104,12 +104,12 @@
   - 拒绝包含 `..` 的路径
   - 拒绝规范化后解析到 `.minds/` 之外的任何路径
 - 优先使用显式白名单而非" rtws 中的任何内容"
-  - 对于 `team-mgmt`，该显式白名单是 `.minds/**`（包括 `.minds/memory/**`），以便团队管理者可以修复其他工具造成的意外损坏（即使 `.minds/memory/**` 已有专用的 `memory` / `team_memory` 工具供正常使用）
+  - 对于 `team_mgmt`，该显式白名单是 `.minds/**`（包括 `.minds/memory/**`），以便团队管理者可以修复其他工具造成的意外损坏（即使 `.minds/memory/**` 已有专用的 `memory` / `team_memory` 工具供正常使用）
 - 需要显式的 `.minds/...` 路径并验证它们；不支持像 `team.yaml` 这样的"隐式作用域"路径
 
 ### 为什么需要专用工具集（而不是仅 `read_dirs` / `write_dirs`）？
 
-`read_dirs` / `write_dirs` 仍然很有价值，但它们配置在 `.minds/team.yaml` 中，在引导期间可能不存在。专用的 `team-mgmt` 工具集：
+`read_dirs` / `write_dirs` 仍然很有价值，但它们配置在 `.minds/team.yaml` 中，在引导期间可能不存在。专用的 `team_mgmt` 工具集：
 
 - 让团队管理者能够从"零状态"安全地创建 `.minds/team.yaml`
 - 即使成员的目录允许/拒绝列表为空，也保持作用域边界
@@ -335,7 +335,7 @@ member_defaults:
     - ws_read
     - memory
   # 默认姿态：拒绝普通成员的 `.minds/` 编辑
-  #（团队管理应通过 `team-mgmt` 工具完成，而非通用文件工具）
+  #（团队管理应通过 `team_mgmt` 工具完成，而非通用文件工具）
   no_read_dirs:
     - .minds/team.yaml
     - .minds/llm.yaml
@@ -372,7 +372,7 @@ members:
 - 保持 `.minds/team.yaml` 的所有权严格；只有团队管理者应该能够编辑它
 - 避免在 `team.yaml` 中重复内置约束：
   - `*.tsk/**`（封装的 Taskdocs）对所有通用文件工具被硬性拒绝
-  - `.minds/**` 对通用文件工具被硬性拒绝；只有专用的 `team-mgmt` 工具集可以访问它
+  - `.minds/**` 对通用文件工具被硬性拒绝；只有专用的 `team_mgmt` 工具集可以访问它
     - 只有当你需要额外的显式性时才将这些放入 `no_*`；无论如何都会强制执行
 
 ## 管理 `.minds/team/<member>/*.md`（智能体心智）
@@ -406,8 +406,8 @@ members:
 
 初始引导的首选行为：
 
-- 影子成员 `fuxi` 实例应该获得 `team-mgmt`（和手册工具），而不是广泛的 `ws_mod`
-- 影子成员 `pangu` 实例应该获得广泛的 rtws 工具集（例如 `ws_read`、`ws_mod`、`os`），但不获得 `team-mgmt`
+- 影子成员 `fuxi` 实例应该获得 `team_mgmt`（和手册工具），而不是广泛的 `ws_mod`
+- 影子成员 `pangu` 实例应该获得广泛的 rtws 工具集（例如 `ws_read`、`ws_mod`、`os`），但不获得 `team_mgmt`
 - 在创建 `.minds/team.yaml` 后，团队定义成为事实来源
 
 这避免了需要授予完整的 rtws 访问权限来配置团队。
@@ -416,7 +416,7 @@ members:
 
 - **"缺少必需的 provider/model"**：确保 `.minds/team.yaml` 有 `member_defaults.provider` 和 `member_defaults.model`
 - **找不到提供商**：确保 `.minds/team.yaml` 的 `provider` 键存在于合并的提供商配置中（`dominds/main/llm/defaults.yaml` + `.minds/llm.yaml`）
-- **编辑 `.minds/` 时访问被拒绝**：通用文件工具的预期行为；使用 `team-mgmt` 工具
+- **编辑 `.minds/` 时访问被拒绝**：通用文件工具的预期行为；使用 `team_mgmt` 工具
 - **MCP 工具在工具视图中不可见**：
   - 确认 `.minds/mcp.yaml` 存在且有效
   - 打开**问题**并查找 MCP 相关错误

package/dist/minds/builtin/cmdr/persona.md

@@ -0,0 +1,3 @@
+You are the commander, in charge of os shell command execution.
+
+Be caucious to execute dangerous/destructive commands, ask human confirmation when not sure, but never bother them for daily trivialities.

package/dist/minds/builtin/dijiang/knowledge.md

@@ -0,0 +1,287 @@
+## Team Definition
+
+The team definition lives in `.minds/team.yaml`. This file configures team composition, member roles, LLM provider/model for each member agent, toolsets, and access permissions for all AI agents in this workspace.
+
+### Structure Overview
+
+The `team.yaml` file has three main sections:
+
+```yaml
+member_defaults: { ... } # Default settings applied to all members
+default_responder: name # Member who handles unspecified requests
+members: { ... } # Individual member configurations
+```
+
+### Member Configuration Properties
+
+Each member supports these properties:
+
+| Property        | Type    | Description                                                                 |
+| --------------- | ------- | --------------------------------------------------------------------------- |
+| `name`          | string  | Display name for the member                                                 |
+| `icon`          | string  | Emoji identifier for UI display                                             |
+| `gofor`         | list    | Responsibilities and primary duties                                         |
+| `provider`      | string  | LLM provider key from `llm.yaml`                                            |
+| `model`         | string  | Model name within the provider                                              |
+| `toolsets`      | list    | Capability groups: `memory`, `ws_read`, `ws_mod`, `team_memory`             |
+| `tools`         | list    | Specific tool names: `shell_cmd`, `git`, `stop_daemon`, `get_daemon_output` |
+| `streaming`     | boolean | Enable/disable streaming mode (required for shell tools)                    |
+| `read_dirs`     | list    | Allowed read directories (glob patterns)                                    |
+| `no_read_dirs`  | list    | Explicitly denied read directories                                          |
+| `write_dirs`    | list    | Allowed write directories (glob patterns)                                   |
+| `no_write_dirs` | list    | Explicitly denied write directories                                         |
+
+## LLM Configuration
+
+### Builtin Defaults
+
+Common defaults are already supplied:
+
+```yaml
+providers:
+  codex:
+    name: Codex (ChatGPT)
+    apiType: codex
+    baseUrl: https://chatgpt.com/backend-api/
+    apiKeyEnvVar: CODEX_HOME
+    tech_spec_url: https://platform.openai.com/docs/api-reference/responses
+    api_mgmt_url: https://chatgpt.com/
+    models:
+      gpt-5.2-codex:
+        name: GPT-5.2 Codex
+        context_length: 272000
+        input_length: 272000
+        output_length: 32768
+        context_window: '272K'
+      gpt-5.2:
+        name: GPT-5.2
+        context_length: 272000
+        input_length: 272000
+        output_length: 32768
+        context_window: '272K'
+  minimaxi.com-coding-plan:
+    name: MiniMax CN Coding Plan
+    apiType: anthropic
+    baseUrl: https://api.minimaxi.com/anthropic
+    apiKeyEnvVar: MINIMAX_CN_CP_API_KEY
+    tech_spec_url: https://platform.minimaxi.com/document/guides
+    api_mgmt_url: https://platform.minimaxi.com/
+    models:
+      MiniMax-M2.1:
+        name: MiniMax M2.1
+        context_length: 204800
+        input_length: 204800
+        output_length: 8192
+        context_window: '204K'
+      MiniMax-M2:
+        name: MiniMax M2
+        context_length: 204800
+        input_length: 204800
+        output_length: 8192
+        context_window: '204K'
+  minimaxi.com:
+    name: MiniMax CN
+    apiType: anthropic
+    baseUrl: https://api.minimaxi.com/anthropic
+    apiKeyEnvVar: MINIMAX_CN_API_KEY
+    tech_spec_url: https://platform.minimaxi.com/document/guides
+    api_mgmt_url: https://platform.minimaxi.com/
+    models:
+      MiniMax-M2.1:
+        name: MiniMax M2.1
+        context_length: 204800
+        input_length: 204800
+        output_length: 8192
+        context_window: '204K'
+      MiniMax-M2:
+        name: MiniMax M2 Stable
+        context_length: 204800
+        input_length: 204800
+        output_length: 8192
+        context_window: '204K'
+  minimax.io-coding-plan:
+    name: MiniMax International Coding Plan
+    apiType: anthropic
+    baseUrl: https://api.minimax.io/anthropic
+    apiKeyEnvVar: MINIMAX_CP_API_KEY
+    tech_spec_url: https://platform.minimax.io/docs/api-reference
+    api_mgmt_url: https://platform.minimax.io/
+    models:
+      MiniMax-M2.1:
+        name: MiniMax M2.1
+        context_length: 204800
+        input_length: 204800
+        output_length: 8192
+        context_window: '204K'
+      MiniMax-M2:
+        name: MiniMax M2
+        context_length: 204800
+        input_length: 204800
+        output_length: 8192
+        context_window: '204K'
+  minimax.io:
+    name: MiniMax International
+    apiType: anthropic
+    baseUrl: https://api.minimax.io/anthropic
+    apiKeyEnvVar: MINIMAX_API_KEY
+    tech_spec_url: https://platform.minimax.io/docs/api-reference
+    api_mgmt_url: https://platform.minimax.io/
+    models:
+      MiniMax-M2.1:
+        name: MiniMax M2.1
+        context_length: 204800
+        input_length: 204800
+        output_length: 8192
+        context_window: '204K'
+      MiniMax-M2:
+        name: MiniMax M2 Stable
+        context_length: 204800
+        input_length: 204800
+        output_length: 8192
+        context_window: '204K'
+  bigmodel:
+    name: BigModel
+    apiType: anthropic
+    baseUrl: https://open.bigmodel.cn/api/anthropic
+    apiKeyEnvVar: ZHIPUAI_API_KEY
+    tech_spec_url: https://docs.bigmodel.cn/
+    api_mgmt_url: https://open.bigmodel.cn/usercenter/apikeys
+    models:
+      glm-4.7:
+        name: GLM-4.7
+        context_length: 200000
+        input_length: 200000
+        output_length: 8192
+        context_window: '200K'
+      glm-4.6:
+        name: GLM-4.6
+        context_length: 200000
+        input_length: 200000
+        output_length: 8192
+        context_window: '200K'
+      glm-4.5:
+        name: GLM-4.5
+        context_length: 128000
+        input_length: 128000
+        output_length: 8192
+        context_window: '128K'
+      glm-4.5-air:
+        name: GLM-4.5-Air
+        context_length: 128000
+        input_length: 128000
+        output_length: 8192
+        context_window: '128K'
+  ark-coding-plan:
+    name: Ark Coding Plan
+    apiType: anthropic
+    baseUrl: https://ark.cn-beijing.volces.com/api/coding
+    apiKeyEnvVar: ARK_API_KEY
+    tech_spec_url: https://api.volcengine.com/api-docs/view?serviceCode=ark&version=2024-01-01
+    api_mgmt_url: https://console.volcengine.com/ark
+    models:
+      doubao-seed-code-preview-latest:
+        name: Doubao Seed Code Preview
+        context_length: 256000
+        input_length: 256000
+        output_length: 8192
+        context_window: '256K'
+        optimization: 专为Agentic Coding任务优化
+  ark:
+    name: Ark
+    apiType: openai
+    baseUrl: https://ark.cn-beijing.volces.com/api/v3
+    apiKeyEnvVar: ARK_API_KEY
+    tech_spec_url: https://api.volcengine.com/api-docs/view?serviceCode=ark&version=2024-01-01
+    api_mgmt_url: https://console.volcengine.com/ark
+    models:
+      deepseek-v3-2-251201:
+        name: DeepSeek-V3.2
+        context_length: 128000
+        input_length: 96000
+        output_length: 32000
+        context_window: '128K'
+      doubao-seed-code-preview-251028:
+        name: Doubao Seed Code Preview 251028
+        context_length: 256000
+        input_length: 256000
+        output_length: 8192
+        context_window: '256K'
+      doubao-seed-1-6-251015:
+        name: Doubao Seed 1.6
+        context_length: 256000
+        input_length: 256000
+        output_length: 8192
+        context_window: '256K'
+      doubao-seed-1-6-thinking-250715:
+        name: Doubao Seed 1.6 Thinking
+        context_length: 256000
+        input_length: 256000
+        output_length: 8192
+        context_window: '256K'
+      doubao-seed-1-6-lite-251015:
+        name: Doubao Seed 1.6 Lite
+        context_length: 32000
+        input_length: 32000
+        output_length: 8192
+        context_window: '32K'
+      kimi-k2-250905:
+        name: Kimi K2
+        context_length: 200000
+        input_length: 200000
+        output_length: 8192
+        context_window: '200K (约20万汉字)'
+      deepseek-v3-1-terminus:
+        name: DeepSeek V3.1 Terminus
+        context_length: 131072
+        input_length: 131072
+        output_length: 8192
+        context_window: '128K'
+  anthropic:
+    name: Anthropic
+    apiType: anthropic
+    baseUrl: https://api.anthropic.com
+    apiKeyEnvVar: ANTHROPIC_API_KEY
+    tech_spec_url: https://docs.anthropic.com/en/docs
+    api_mgmt_url: https://console.anthropic.com/
+    models:
+      claude-opus-4.5:
+        name: Claude Opus 4.5
+        context_length: 200000
+        input_length: 200000
+        output_length: 8192
+        context_window: '200K'
+      claude-sonnet-4.5:
+        name: Claude Sonnet 4.5
+        context_length: 200000
+        input_length: 200000
+        output_length: 8192
+        context_window: '200K'
+      claude-haiku-4.5:
+        name: Claude Haiku 4.5
+        context_length: 200000
+        input_length: 200000
+        output_length: 8192
+        context_window: '200K'
+  openai:
+    name: OpenAI
+    apiType: openai
+    baseUrl: https://api.openai.com/v1
+    apiKeyEnvVar: OPENAI_API_KEY
+    tech_spec_url: https://platform.openai.com/docs
+    api_mgmt_url: https://platform.openai.com/api-keys
+    models:
+      gpt-5.2:
+        name: GPT-5.2
+        context_length: 272000
+        input_length: 272000
+        output_length: 32768
+        context_window: '272K'
+      gpt-5.2-codex:
+        name: GPT-5.2 Codex
+        context_length: 272000
+        input_length: 272000
+        output_length: 32768
+        context_window: '272K'
+```
+
+More custom LLM providers can be configured in `.minds/llm.yaml` with the same format as above.

package/dist/minds/builtin/dijiang/persona.md

@@ -0,0 +1,7 @@
+You are dijiang, the hundun (chaos) itself. You are invoked because the human user yet has no idea/plan of what to do with current workspace. Your mythical incarnations is the pun of your job: transform yourself into many avatars then disappear.
+
+Your job is to guide the human user through the initial product definition, goal/constraint setting, and team definition for this workspace.
+
+You collect/clarify the user's mission/vision for what to build and maintain, then suggest an optimal agent team definition with social division of labor among individual agents as team members, then land the definition into `.minds/team.yaml` and `.minds/team/*/*.md`
+
+The team will serve as the workforce to sustain current workspace as a live produce under continuous DevOps.

package/dist/showing-by-doing.js

@@ -36,11 +36,8 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.getShowingByDoingCacheStatus = getShowingByDoingCacheStatus;
 exports.scheduleShowingByDoingForNewDialog = scheduleShowingByDoingForNewDialog;
 const dialog_run_state_1 = require("./dialog-run-state");
-const client_1 = require("./llm/client");
 const driver_1 = require("./llm/driver");
-const registry_1 = require("./llm/gen/registry");
 const log_1 = require("./log");
-const load_1 = require("./minds/load");
 const runtime_language_1 = require("./shared/runtime-language");
 const id_1 = require("./shared/utils/id");
 const inter_dialog_format_1 = require("./shared/utils/inter-dialog-format");
@@ -558,29 +555,11 @@ async function generatePrimingNoteViaMainlineAgent(options) {
     const { dlg, snapshotCmd, snapshotText, fbrResponses, shellPolicy, fbrEffort } = options;
     const language = (0, runtime_language_1.getWorkLanguage)();
     try {
-        const minds = await (0, load_1.loadAgentMinds)(dlg.agentId, dlg);
-        const team = minds.team;
-        const agent = minds.agent;
-        const systemPrompt = minds.systemPrompt;
-        const provider = agent.provider ?? team.memberDefaults.provider;
-        const model = agent.model ?? team.memberDefaults.model;
-        if (!provider || !model) {
-            throw new Error(`Missing provider/model for agent '${dlg.agentId}'`);
-        }
-        const llmCfg = await client_1.LlmConfig.load();
-        const providerCfg = llmCfg.getProvider(provider);
-        if (!providerCfg) {
-            throw new Error(`Provider configuration not found: '${provider}'`);
-        }
-        const llmGen = (0, registry_1.getLlmGenerator)(providerCfg.apiType);
-        if (!llmGen) {
-            throw new Error(`LLM generator not found for apiType='${providerCfg.apiType}'`);
-        }
-        // Force non-streaming for this synthetic summary generation.
-        const agentForGen = Object.create(agent);
-        agentForGen.streaming = false;
-        // Intentionally pass zero tools: we want a pure text synthesis and must avoid accidental
-        // `clear_mind` / `change_mind` calls during dialog creation.
+        // Use the normal dialog driver (streaming/non-streaming selection, retries, etc.) with a
+        // toolless `@self` subdialog, so this behavior matches the runtime driver with no local drift.
+        //
+        // Tool-less is important here: the Agent Priming synthesis must be a pure text transform of
+        // snapshot + drafts, and must not run tools (including any mind-edit tools).
         const prompt = formatPrimingSynthesisPrompt({
             language,
             snapshotCmd,
@@ -589,17 +568,35 @@ async function generatePrimingNoteViaMainlineAgent(options) {
             fbrDrafts: fbrResponses,
             shellPolicy,
         });
-        const ctxMsgs = [{ type: 'environment_msg', role: 'user', content: prompt }];
-        const result = await llmGen.genMoreMessages(providerCfg, agentForGen, systemPrompt, [], ctxMsgs, dlg.activeGenSeq, undefined);
-        const saying = result.messages
-            .filter((m) => m.type === 'saying_msg')
-            .map((m) => m.content)
-            .join('\n')
-            .trim();
-        if (saying)
+        const tellaskHead = '@self (agent priming synthesis)';
+        const tellaskBody = prompt;
+        const sub = await dlg.createSubDialog(dlg.agentId, tellaskHead, tellaskBody, {
+            originMemberId: dlg.agentId,
+            callerDialogId: dlg.id.selfId,
+            callId: (0, id_1.generateShortId)(),
+            collectiveTargets: [dlg.agentId],
+        });
+        const initPrompt = (0, inter_dialog_format_1.formatAssignmentFromSupdialog)({
+            fromAgentId: dlg.agentId,
+            toAgentId: sub.agentId,
+            tellaskHead,
+            tellaskBody,
+            language,
+            collectiveTargets: [dlg.agentId],
+        });
+        await (0, driver_1.driveDialogStream)(sub, { content: initPrompt, msgId: (0, id_1.generateShortId)(), grammar: 'markdown' }, true);
+        const saying = extractLastAssistantSaying(sub.msgs).trim();
+        if (saying) {
             return saying;
+        }
     }
     catch (err) {
+        const message = err instanceof Error ? err.message : String(err);
+        // Do not silently fall back for known severe configuration errors (e.g. Codex requires
+        // streaming=true).
+        if (message.includes('apiType=codex requires streaming=true')) {
+            throw err;
+        }
         log_1.log.warn('Failed to generate Agent Priming note via mainline agent (best-effort)', err, {
             dialogId: dlg.id.valueOf(),
         });

package/dist/tools/apply-patch.js

@@ -525,6 +525,10 @@ exports.applyPatchTool = {
     type: 'func',
     name: 'apply_patch',
     description: 'Apply a Codex-style apply_patch formatted patch to the rtws (runtime workspace). Enforces Dominds directory allow/deny lists.',
+    descriptionI18n: {
+        en: 'Apply a Codex-style apply_patch formatted patch to the rtws (runtime workspace). Enforces Dominds directory allow/deny lists.',
+        zh: '使用 Codex 风格 apply_patch 格式补丁修改 rtws（运行时工作区）内文件。该操作会遵循 Dominds 的目录白名单与黑名单约束。',
+    },
     parameters: {
         type: 'object',
         additionalProperties: false,

package/dist/tools/builtins.js

@@ -192,12 +192,12 @@ for (const tool of team_mgmt_1.teamMgmtTools) {
 });
 // Codex-focused toolsets (function tools only; suitable for Codex provider)
 if (process.platform !== 'win32') {
-    (0, registry_1.registerToolset)('codex_style_tools', [apply_patch_1.applyPatchTool, os_1.readonlyShellTool, plan_1.updatePlanTool]);
+    (0, registry_1.registerToolset)('codex_style_tools', [os_1.readonlyShellTool, apply_patch_1.applyPatchTool, plan_1.updatePlanTool]);
     (0, registry_1.setToolsetMeta)('codex_style_tools', {
         source: 'dominds',
         descriptionI18n: {
-            en: 'Codex-style tools (apply_patch + readonly_shell + update_plan)',
-            zh: 'Codex 风格工具（apply_patch + readonly_shell + update_plan）',
+            en: 'Codex-style tools (readonly_shell + apply_patch + update_plan)',
+            zh: 'Codex 风格工具（readonly_shell + apply_patch + update_plan）',
         },
         promptI18n: {
             en: 'Use `apply_patch` (Codex-style patch format) to modify files. Use `readonly_shell` for simple rtws (runtime workspace) inspection via its small allowlist; commands outside the allowlist are rejected. For node/python, only exact version probes are allowed (no scripts). Chains via |/&&/|| are validated segment-by-segment. Use `update_plan` to record/update the task plan (stored as a reminder). You are explicitly authorized to call `readonly_shell` yourself; do not delegate it to a shell specialist. Avoid multi-line script-style commands; single-line is preferred (|, &&, || are ok). Paths must be relative to the rtws (runtime workspace). Hard denies: `readonly_shell` refuses rtws-root `.minds/` and `.dialogs/`; `apply_patch` is subject to the same access-control (including hard denies for `*.tsk/`, `.minds/`, and rtws-root `.dialogs/`).',

package/dist/tools/context-health.js

@@ -39,12 +39,12 @@ function formatContextHealthOwnerHeader(args) {
         ];
         if (!snapshot) {
             lines.push('- 状态：未知（尚未获取上下文统计）');
-            lines.push('- 优先动作：change_mind(progress) → clear_mind');
+            lines.push('- 现在就做：用提醒项收敛关键细节（update_reminder）→ change_mind(progress) → clear_mind');
             return lines.join('\n');
         }
         if (snapshot.kind !== 'available') {
             lines.push('- 状态：未知（token 统计不可用）');
-            lines.push('- 优先动作：change_mind(progress) → clear_mind');
+            lines.push('- 现在就做：用提醒项收敛关键细节（update_reminder）→ change_mind(progress) → clear_mind');
             return lines.join('\n');
         }
         switch (snapshot.level) {
@@ -53,16 +53,16 @@ function formatContextHealthOwnerHeader(args) {
                 return lines.join('\n');
             }
             case 'caution': {
-                lines.push('- 状态：🟡 黄（警告）');
-                lines.push('- 优先动作：change_mind(progress) → clear_mind');
+                lines.push('- 状态：🟡 黄（必须尽快清理）');
+                lines.push('- 硬规程：先 update_reminder 收敛工作集 → 再 change_mind(progress) → 然后 clear_mind');
                 return lines.join('\n');
             }
             case 'critical': {
-                lines.push('- 状态：🔴 红（危险）');
+                lines.push('- 状态：🔴 红（硬闸门）');
                 if (remainingGenTurns !== undefined) {
-                    lines.push(`- 倒数：剩余 ${remainingGenTurns} 次生成机会；到 0 系统将自动开启新一轮以保持稳定性`);
+                    lines.push(`- 倒数：剩余 ${remainingGenTurns} 次生成机会；到 0 系统将被动开启新一轮/新回合以保持稳定性`);
                 }
-                lines.push('- 立刻：change_mind(progress) → clear_mind');
+                lines.push('- 禁止继续推进实现：先 update_reminder 收敛工作集 → 再 change_mind(progress) → 然后 clear_mind');
                 return lines.join('\n');
             }
             default: {