@deepseek-ai/dsh-tool-terminal 0.1.1-rc.2 → 0.1.2-alpha.3

package/README.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/terminal/tool-terminal/README.md
-README.md: f6fd4020d423425d7b20ad6bc21b4e009cf25c85
-README.zh.md: dca0ce9280fc7a7d748e90a5c553c0e712e2453d
+README.md: 0ccdb451e829f24cffb0f2c89d4d70ffe9d40c68
+README.zh.md: f7852bd8185086e4f0456d1c79821e71ef31472c

package/README.md

@@ -1,20 +1,118 @@
+---
+description: "Six model-facing persistent terminal tools with owner isolation, bounded results, and optional background sends for agents that need cross-call terminal state."
+kind: "package-reference"
+---
+
 # @deepseek-ai/dsh-tool-terminal
 
 English | [中文](README.zh.md)
 
-Six model-facing tools over `ctx.terminals`: `terminal_open`, `terminal_send`, `terminal_read`, `terminal_signal`, `terminal_close`, and `terminal_list`. Every operation requires the exact initiating `Agent`, so a model cannot address another agent's terminal even if it learns the id.
+## Summary
+
+`dsh-tool-terminal` gives the model six tools over persistent terminal sessions: `terminal_open`, `terminal_send`, `terminal_read`, `terminal_signal`, `terminal_close`, and `terminal_list`. Every call is fenced to the exact agent that opened the session, so a model cannot operate another agent's terminal even if it learns the id. Sends run in the foreground (returning bounded output with a wait reason) or in the background through the jobs service (returning a job id collected with `job_output` and stopped with `job_kill`). Results are capped by `maxResultBytes` and stay in session history until compaction. A short guidance section tells the model to prefer one-shot tools unless a terminal's persistent state or interactive stdin is genuinely needed.
+
+## Table of Contents
+
+- [Use this package](#use-this-package)
+- [Understand the implementation](#understand-the-implementation)
+- [Further Exploration](#further-exploration)
+- [Model Experience](#model-experience)
+- [Known Limitations and Deferred Work](#known-limitations-and-deferred-work)
+- [Dev Note](#dev-note)
+
+-----
+
+<a id="use-this-package"></a>
+## Use this package
+
+Enable these tools when the composition mounts a terminal backend and the model should be able to use terminal state across calls — stepping a debugger, exploring in a REPL, or returning to a shell after interrupting a foreground command. The guidance section steers the model toward the one-shot bash, read, write, and edit tools for bounded operations.
+
+### The six tools
+
+| Tool | What it does | Result |
+|---|---|---|
+| `terminal_open` | Creates an owner-scoped session from a backend type | Session id, name, type, pid, status, and bounded startup output |
+| `terminal_send` | Writes text, optionally submitting Enter, and waits for readiness — or starts a background job | Bounded output plus wait and session status, or a job id |
+| `terminal_read` | Reads a bounded page of retained output without sending input | Text with line pagination metadata |
+| `terminal_signal` | Delivers one allowed signal to the foreground process group | `delivered` plus the target process group id |
+| `terminal_close` | Closes a session and waits for its process tree to end | Closed or already-closing outcome |
+| `terminal_list` | Lists the caller's live sessions | Owner-scoped session summaries |
+
+### Composition
+
+```yaml
+- name: '@deepseek-ai/dsh-terminal'
+- name: '@deepseek-ai/dsh-terminal-bash'
+- name: '@deepseek-ai/dsh-tool-terminal'
+```
+
+The tools need `ctx.terminals` — a backend must be mounted — and the system-prompt service for the guidance section. Background sends additionally require the jobs service and its model-facing controller (`@deepseek-ai/dsh-tool-jobs`).
+
+### Configuration
+
+| Field | Default | Meaning |
+|---|---|---|
+| `enableRunInBackground` | `true` | Expose and accept `run_in_background`; `false` removes the schema field and rejects the argument |
+| `maxResultBytes` | `262144` | UTF-8 cap (minimum `64`) for each complete terminal result after wait, session, pagination, truncation, and job-status metadata |
+
+The generated [configuration catalog](../../../docs/config-catalog.md#deepseek-aidsh-tool-terminal) and [tool catalog](../../../docs/tool-catalog.md#deepseek-aidsh-tool-terminal) are the exhaustive sources for config fields and schemas.
+
+### Background sends
+
+`terminal_send(run_in_background: true)` returns a job id immediately instead of waiting. The job is collected with `job_output`, which waits and reads incremental output, and stopped with `job_kill`, which delivers a real `SIGINT` to the foreground process group. Background mode fails before writing input when the jobs surface is absent.
+
+### Observable outcomes and failures
+
+A foreground send returns the terminal's new output plus `wait: <reason>` and the session status; `session_exit` means the top-level shell exited, while `inferred_idle` or `timeout` never proves the foreground command exited. Opening a session with an unregistered backend type fails. Results larger than `maxResultBytes` are truncated at a UTF-8 boundary with a marker.
+
+-----
 
-`terminal_send(run_in_background: true)` reuses `ctx.jobs`; job preflight and the PTY service's exclusive per-session send reservation occur before the job id is returned, completion is collected with `job_output`, and `job_kill` delivers `SIGINT` to the foreground process group. Foreground sends use terminal call/result cards. Background sends use a generic execute card; open, read, signal, close, and list use generic `execute`, `read`, `execute`, `delete`, and `read` cards respectively. None declares source locations.
+<a id="understand-the-implementation"></a>
+## Understand the implementation
 
-## Config
+<details>
+<summary>Implementation internals — click to expand</summary>
 
-| key | default | meaning |
-|---|---:|---|
-| `enableRunInBackground` | `true` | expose and accept `run_in_background`; false omits the schema field and rejects a forced undeclared argument |
-| `maxResultBytes` | `262144` | UTF-8 cap (minimum `64`) for each complete terminal result or PTY job output after wait, session, pagination, truncation, and task-status metadata |
+This section explains the design behind the tools and points at the code that realizes it; the observable behavior is covered in [Use this package](#use-this-package).
 
-Both values are validated at load. The minimum result cap keeps every registry-issued session or job id visible in its creation acknowledgement. When a result exceeds `maxResultBytes`, rendering reserves space for control metadata and a truncation marker when they fit; cuts preserve UTF-8 boundaries. Each terminal definition's final-content callback applies the same cap after normalized pre-, around-, and post-execute policy failures, denials, short-circuits, replacements, or blocks; a structured multi-block policy result retains its shape.
+### Design concept
 
+The package is a thin adapter: the six tools forward to `ctx.terminals` with the executing agent as the owner, and the presentation layer renders bounded results. Background sends register the in-flight operation on `ctx.jobs` so the generic job surface owns waiting, incremental reads, and `SIGINT` delivery.
+
+### Source map
+
+| File | Role |
+|---|---|
+| [`src/index.ts`](src/index.ts) | Six tool definitions, schemas, guidance section, background-job integration |
+| [`src/render.ts`](src/render.ts) | Result rendering and the complete-result UTF-8 cap |
+
+### Result bounding
+
+Every terminal-owned single-text result is capped by `maxResultBytes` after normalized tool or pipeline errors, policy denials and short-circuits, replacements and blocks, and generic job-status text; cuts preserve UTF-8 boundaries and reserve room for a truncation marker. Structured multi-block policy results retain their shape. The minimum cap of 64 bytes keeps every registry-issued session or job id visible in its creation acknowledgement.
+
+### UI render intents
+
+Foreground sends use terminal call and result cards; background sends and the other five tools use generic `execute`, `read`, or `delete` cards. None of the tools emits source locations.
+
+</details>
+
+-----
+
+<a id="further-exploration"></a>
+## Further Exploration
+
+Read these pages when the package-level contract is not enough. They move from the generated schemas to the service contract, the backend, and the background-job surface.
+
+- [Tool catalog](../../../docs/tool-catalog.md#deepseek-aidsh-tool-terminal) — the six generated schemas and result shapes.
+- [Terminal subsystem reference](../../../docs/subsystems/terminal.md) — the service contract and shared types behind the tools.
+- [terminal service](../terminal/README.md) — session operations, owner fencing, and cleanup semantics.
+- [terminal-bash backend](../terminal-bash/README.md) — the shipped shell backend that provides sessions.
+- [jobs package map](../../jobs/README.md) — the background-job surface that collects and kills background sends.
+- [Persistent PTY Agent Note](../../../.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.md) — the capability design and deferred boundaries.
+
+-----
+
+<a id="model-experience"></a>
 ## Model Experience
 
 ### System prompt
@@ -55,11 +153,11 @@ Prefix-stable while tool visibility and definitions are unchanged.
 
 #### What the model sees
 
-Spawn returns the id and bounded MOTD. Send/read return bounded terminal text plus readiness/history markers. Background mode returns a generic job id. Every terminal-owned or policy-produced single-text result is capped by `maxResultBytes` after normalized tool or pipeline errors, denials, short-circuits, replacements, blocks, and generic job status text. Structured multi-block policy results retain their shape. Results remain in session history until compaction; incremental task reads do not repeat consumed output. Programmatic callers receive typed session snapshots, bounded provider read/send DTOs, signal and close outcomes, or `{ kind: "background", jobId }`; Native rendering applies the presentation cap above.
+Spawn returns the id and bounded startup output. Send and read return bounded terminal text plus readiness and history markers. Background mode returns a generic job id. Every terminal-owned single-text result is capped by `maxResultBytes`; results remain in session history until compaction, and incremental task reads do not repeat consumed output.
 
 #### Token effect
 
-Terminal-owned and policy-produced single-text results are data-dependent and bounded by `maxResultBytes`; a policy that deliberately substitutes structured multi-block content owns that content's bound. Each returned result remains in history until compaction.
+Terminal-owned results are data-dependent and bounded by `maxResultBytes`; each returned result stays in history until compaction.
 
 #### KV Cache effect
 
@@ -67,5 +165,20 @@ Append-only; new results follow the reusable request prefix.
 
 ## Known Limitations and Deferred Work
 
-- No named key sequence, TUI, BEL, resize, auto-start, or cross-agent sharing schema is exposed.
-- Background mode requires both `@deepseek-ai/dsh-jobs` and its model-facing controller.
+<a id="known-limitations-and-deferred-work"></a>
+
+
+These limits define the model-facing surface that is absent. They are current package constraints, not a task backlog.
+
+- **No TUI or key-sequence surface** — named key sequences, full-screen TUI interaction, BEL, resize, and auto-start are not exposed in any schema.
+- **Background mode requires the jobs surface** — `run_in_background` needs both `@deepseek-ai/dsh-jobs` and its model-facing controller (`@deepseek-ai/dsh-tool-jobs`); without them the argument is rejected.
+
+<a id="dev-note"></a>
+### Dev Note
+
+<details>
+<summary>Working context for maintainers — click to expand</summary>
+
+None.
+
+</details>

package/README.zh.md

@@ -1,25 +1,123 @@
+---
+description: "面向需要跨调用终端状态的 agent 的 6 个持久终端工具，带所有者隔离、有界结果与可选后台发送。"
+kind: "package-reference"
+---
+
 # @deepseek-ai/dsh-tool-terminal
 
 [English](README.md) | 中文
 
-基于 `ctx.terminals` 提供 6 个面向模型的工具：`terminal_open`、`terminal_send`、`terminal_read`、`terminal_signal`、`terminal_close` 和 `terminal_list`。每项操作都要求提供完全相同的发起 `Agent`，因此即使模型获知另一个 agent（智能体）的 id，也无法操作其终端。
+## 概述
+
+`dsh-tool-terminal` 基于持久终端会话为模型提供 6 个工具：`terminal_open`、`terminal_send`、`terminal_read`、`terminal_signal`、`terminal_close` 与 `terminal_list`。每次调用都被限制在打开该会话的那个确切 agent（智能体）内，因此即使模型获知另一个 agent 的 id，也无法操作其终端。发送可以前台运行（返回带等待原因的有界输出），也可以通过任务服务后台运行（返回 job id，用 `job_output` 收集、用 `job_kill` 停止）。结果受 `maxResultBytes` 限制，并保留在会话历史中直到压缩（compaction）。一段简短指引会告诉模型：除非确实需要终端的持久状态或交互式 stdin，否则优先使用单次工具。
+
+## 目录
+
+- [使用本包](#use-this-package)
+- [理解实现](#understand-the-implementation)
+- [进一步探索](#further-exploration)
+- [模型体验](#model-experience)
+- [已知限制与延期工作](#known-limitations-and-deferred-work)
+- [开发备注](#dev-note)
+
+-----
+
+<a id="use-this-package"></a>
+## 使用本包
+
+当组合挂载了终端后端、且模型应当能跨调用使用终端状态时启用这些工具——逐步调试 gdb、在 REPL 中探索，或中断前台命令后回到 shell。指引章节会引导模型对确有界操作使用单次 bash、read、write 与 edit 工具。
+
+### 六个工具
+
+| 工具 | 作用 | 结果 |
+|---|---|---|
+| `terminal_open` | 按后端类型创建限定所有者范围的会话 | 会话 id、名称、类型、pid、状态与有界启动输出 |
+| `terminal_send` | 写入文本，可选地提交 Enter，并等待就绪——或启动后台任务 | 有界输出加等待与会话状态，或一个 job id |
+| `terminal_read` | 不发送输入，读取一页有界保留输出 | 带行分页元数据的文本 |
+| `terminal_signal` | 向前台进程组投递一个允许的信号 | `delivered` 加目标进程组 id |
+| `terminal_close` | 关闭会话并等待其进程树结束 | 已关闭或正在关闭的结果 |
+| `terminal_list` | 列出调用方的活跃会话 | 限定所有者范围的会话摘要 |
+
+### 组合方式
+
+```yaml
+- name: '@deepseek-ai/dsh-terminal'
+- name: '@deepseek-ai/dsh-terminal-bash'
+- name: '@deepseek-ai/dsh-tool-terminal'
+```
+
+工具需要 `ctx.terminals`——必须挂载一个后端——以及用于指引章节的系统提示词服务。后台发送还额外要求任务服务及其面向模型的控制器（`@deepseek-ai/dsh-tool-jobs`）。
+
+### 配置
+
+| 字段 | 默认值 | 含义 |
+|---|---|---|
+| `enableRunInBackground` | `true` | 公开并接受 `run_in_background`；设为 `false` 时移除 schema 字段并拒绝该参数 |
+| `maxResultBytes` | `262144` | 每个完整终端结果的 UTF-8 上限（最小值 `64`）；在等待、会话、分页、截断与任务状态元数据全部加入后计算 |
+
+生成的[配置目录](../../../docs/config-catalog.zh.md#deepseek-aidsh-tool-terminal)与[工具目录](../../../docs/tool-catalog.zh.md#deepseek-aidsh-tool-terminal)是配置字段与 schema 的穷尽式真源。
+
+### 后台发送
+
+`terminal_send(run_in_background: true)` 立即返回 job id，而不是等待。任务用 `job_output` 收集——它会等待并读取增量输出——用 `job_kill` 停止，后者向前台进程组投递真正的 `SIGINT`。缺少任务接口面时，后台模式会在写入输入之前失败。
+
+### 可观察结果与失败
+
+前台发送返回终端的新输出以及 `wait: <原因>` 与会话状态；`session_exit` 表示顶层 shell 已退出，而 `inferred_idle` 或 `timeout` 绝不证明前台命令已退出。用未注册的后端类型打开会话会失败。大于 `maxResultBytes` 的结果会在 UTF-8 边界处截断并附标记。
+
+-----
 
-`terminal_send(run_in_background: true)` 会复用 `ctx.jobs`；任务预检和 PTY 服务对每个会话的独占发送预留都发生在返回 job id 之前。系统通过 `job_output` 收集完成结果，`job_kill` 则向前台进程组发送 `SIGINT`。前台发送使用终端调用／结果卡片。后台发送使用通用执行卡片；打开、读取、发送信号、关闭和列出操作则分别使用通用 `execute`、`read`、`execute`、`delete` 和 `read` 卡片。所有操作都不声明源位置。
+<a id="understand-the-implementation"></a>
+## 理解实现
 
-## 配置
+<details>
+<summary>实现细节——点击展开</summary>
 
-| 键 | 默认值 | 含义 |
-|---|---:|---|
-| `enableRunInBackground` | `true` | 公开并接受 `run_in_background`；设为 false 时，schema 会省略该字段，并拒绝强行传入未声明的参数 |
-| `maxResultBytes` | `262144` | 每个完整终端结果或 PTY 任务输出的 UTF-8 上限（最小值 `64`）；在等待、会话、分页、截断和任务状态元数据全部加入后计算 |
+本节解释工具背后的设计并指出实现它们的代码位置；可观察行为已在[使用本包](#use-this-package)中说明。
 
-两个值都会在加载时验证。最小结果上限可保证注册表签发的每个会话或 job id 都能出现在创建确认中。结果超过 `maxResultBytes` 时，只要空间允许，渲染会为控制元数据和截断标记预留空间；截断会保留 UTF-8 边界。每个终端定义的最终内容回调都会应用同一个上限，涵盖经过规范化的 pre-execute、around-execute 与 post-execute 策略失败、拒绝、短路、替换或阻止；结构化的多块策略结果保留其结构。
+### 设计理念
 
+本包是薄适配层：6 个工具以执行 agent 作为所有者转发到 `ctx.terminals`，呈现层渲染有界结果。后台发送把在途操作注册到 `ctx.jobs`，由通用任务接口面负责等待、增量读取与 `SIGINT` 投递。
+
+### 源码地图
+
+| 文件 | 职责 |
+|---|---|
+| [`src/index.ts`](src/index.ts) | 6 个工具定义、schema、指引章节、后台任务集成 |
+| [`src/render.ts`](src/render.ts) | 结果渲染与完整结果的 UTF-8 上限 |
+
+### 结果上限
+
+每个终端自身的单文本结果都会在规范化后的工具或流水线错误、策略拒绝与短路、替换与阻止、以及通用任务状态文本之后，受 `maxResultBytes` 限制；截断保留 UTF-8 边界并为截断标记预留空间。结构化的多块策略结果保留其结构。64 字节的最小上限保证注册表签发的每个会话或 job id 都出现在创建确认中。
+
+### UI 呈现意图
+
+前台发送使用终端调用与结果卡片；后台发送与其他 5 个工具使用通用 `execute`、`read` 或 `delete` 卡片。所有工具都不输出源位置。
+
+</details>
+
+-----
+
+<a id="further-exploration"></a>
+## 进一步探索
+
+当包级约定不够用时阅读以下页面。它们从生成的 schema 进入服务约定、后端与后台任务接口面。
+
+- [工具目录](../../../docs/tool-catalog.zh.md#deepseek-aidsh-tool-terminal)——6 个生成的 schema 与结果形态。
+- [终端子系统参考](../../../docs/subsystems/terminal.zh.md)——工具背后的服务约定与共享类型。
+- [terminal 服务](../terminal/README.zh.md)——会话操作、所有者限制与清理语义。
+- [terminal-bash 后端](../terminal-bash/README.zh.md)——提供会话的随附 shell 后端。
+- [jobs 包映射](../../jobs/README.zh.md)——收集与停止后台发送的后台任务接口面。
+- [持久 PTY Agent Note](../../../.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.zh.md)——能力设计与暂缓边界。
+
+-----
+
+<a id="model-experience"></a>
 ## 模型体验
 
 ### 系统提示词
 
-#### 模型看到的内容
+#### 模型看到什么
 
 该插件贡献以下固定指引章节：
 
@@ -35,11 +133,11 @@ Use a terminal session only when work needs persistent terminal state or interac
 
 #### KV Cache 影响
 
-注册范围和指引文本不变时，前缀保持稳定。
+注册范围与指引文本不变时，前缀保持稳定。
 
 ### 工具 schema
 
-#### 模型看到的内容
+#### 模型看到什么
 
 6 个生成的 schema 列在 [`dsh-tool-terminal` 目录章节](../../../docs/tool-catalog.zh.md#deepseek-aidsh-tool-terminal)中。此插件活跃时，请求中会包含它们的固定 schema token；按 agent 范围过滤工具时可能隐藏这些 schema。
 
@@ -53,19 +151,34 @@ Use a terminal session only when work needs persistent terminal state or interac
 
 ### 工具结果与任务上下文
 
-#### 模型看到的内容
+#### 模型看到什么
 
-spawn 会返回 id 和有界 MOTD。发送／读取会返回有界终端文本以及就绪／历史标记。后台模式返回通用 job id。所有终端自身或策略产生的单文本结果，在经过规范化的工具或流水线错误、拒绝、短路、替换、阻止与通用任务状态文本之后，都受 `maxResultBytes` 限制。结构化的多块策略结果保留其结构。结果会保留在会话历史中直到压缩（compaction）；增量任务读取不会重复已经消费的输出。编程调用方会收到带类型的会话快照、有界的提供方读取／发送 DTO、信号与关闭结果，或 `{ kind: "background", jobId }`；Native 渲染会应用上述呈现上限。
+spawn 返回 id 与有界启动输出。发送与读取返回有界终端文本以及就绪与历史标记。后台模式返回通用 job id。每个终端自身的单文本结果都受 `maxResultBytes` 限制；结果保留在会话历史中直到压缩，增量任务读取不会重复已经消费的输出。
 
 #### Token 影响
 
-终端自身与策略产生的单文本结果随数据变化，并受 `maxResultBytes` 限制；如果策略有意替换为结构化多块内容，则由该策略负责限制内容。每个返回结果都会保留在历史中直到压缩。
+终端自身的结果随数据变化，并受 `maxResultBytes` 限制；每个返回结果都保留在历史中直到压缩。
 
 #### KV Cache 影响
 
 仅追加；新结果位于可复用请求前缀之后。
 
-## 已知限制与暂缓事项
+## 已知限制与延期工作
+
+<a id="known-limitations-and-deferred-work"></a>
+
+
+这些限制说明缺失的面向模型接口面。它们是当前包约束，不是任务积压。
+
+- **没有 TUI 或按键序列接口面**——具名按键序列、全屏 TUI 交互、BEL、调整大小与自动启动均未出现在任何 schema 中。
+- **后台模式要求任务接口面**——`run_in_background` 同时需要 `@deepseek-ai/dsh-jobs` 及其面向模型的控制器（`@deepseek-ai/dsh-tool-jobs`）；缺少时会拒绝该参数。
+
+<a id="dev-note"></a>
+### 开发备注
+
+<details>
+<summary>维护者的工作上下文——点击展开</summary>
+
+无。
 
-- 不公开具名按键序列、TUI、BEL、调整大小、自动启动或跨 agent 共享 schema。
-- 后台模式同时依赖 `@deepseek-ai/dsh-jobs` 及其面向模型的控制器。
+</details>

package/lib/index.js

@@ -222,7 +222,7 @@ function apply(ctx, config = {}) {
 	};
 	ctx.systemPrompt.section({
 		name: "tool:pty",
-		order: 106,
+		order: ctx.systemPrompt.getSectionOrder("TOOL_PTY"),
 		text: "Use a terminal session only when work needs persistent terminal state or interactive stdin; prefer shell/read/write/edit for bounded one-shot operations. Track every terminal session id and close sessions that no longer matter. An inferred_idle or timeout result does not prove the foreground command exited."
 	});
 	ctx.tools.register(defineTool({

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@deepseek-ai/dsh-tool-terminal",
   "description": "Six model-facing persistent PTY tools with owner isolation and generic background-job integration",
-  "version": "0.1.1-rc.2",
+  "version": "0.1.2-alpha.3",
   "publishConfig": {
     "access": "public"
   },
@@ -32,37 +32,38 @@
   ],
   "license": "MIT",
   "dependencies": {
-    "@deepseek-ai/schemastery": "^3.18.1"
+    "@deepseek-ai/schemastery": "^3.18.2"
   },
   "peerDependencies": {
-    "@deepseek-ai/dsh-agent": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-invariants": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-llm": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-terminal": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-output-retention": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-jobs": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-system-prompt": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-tools": "^0.1.1-rc.2",
-    "@deepseek-ai/cordis": "^4.0.1"
+    "@deepseek-ai/dsh-agent": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-invariants": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-terminal": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-llm": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-output-retention": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-system-prompt": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-jobs": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-tools": "^0.1.2-alpha.3",
+    "@deepseek-ai/cordis": "^4.0.2"
   },
   "devDependencies": {
-    "@deepseek-ai/cordis-plugin-include": "^1.0.6",
-    "@deepseek-ai/cordis-plugin-loader": "^1.0.2",
-    "@deepseek-ai/dsh-agent": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-invariants": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-llm": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-terminal": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-terminal-bash": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-output-retention": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-sandbox": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-sandbox-policy": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-session": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-subprocess-local": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-system-prompt": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-jobs": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-jobs-local": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-tool-jobs": "^0.1.1-rc.2",
-    "@deepseek-ai/dsh-tools": "^0.1.1-rc.2",
-    "@deepseek-ai/cordis": "^4.0.1"
+    "@deepseek-ai/cordis-plugin-loader": "^1.0.3",
+    "@deepseek-ai/cordis-plugin-include": "^1.0.7",
+    "@deepseek-ai/dsh-agent": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-invariants": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-llm": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-terminal-bash": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-sandbox": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-output-retention": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-terminal": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-session": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-subprocess-local": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-system-prompt": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-jobs": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-jobs-local": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-tool-jobs": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-tools": "^0.1.2-alpha.3",
+    "@deepseek-ai/cordis": "^4.0.2",
+    "@deepseek-ai/dsh-session-projection": "^0.1.2-alpha.3",
+    "@deepseek-ai/dsh-sandbox-policy": "^0.1.2-alpha.3"
   }
 }