@dsh-cc/preset-cc 0.7.1-rc.1

package/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

package/README.i18n.yaml

@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# 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 check:readme --write
+README.md: 915baac2d78d1412da3ef374df2d2c09c91270d9
+README.zh.md: 582c9ec91057a4e4eb77f8e5bb51b4d0032b4505

package/README.md

@@ -0,0 +1,45 @@
+# @dsh-cc/preset-cc
+
+English | [中文](README.zh.md)
+
+The **CC 模式** (CC mode) agent preset for the DeepSeek Harness (dsh): the fifth agent preset alongside the stock `standard`, `minimal`, and the other built-ins. It composes the Claude Code surface — Claude hooks, on-disk CC plugin directories, slash commands, memory, CC output styles, and CC-style WebFetch — on top of the complete standard preset. This package is **composition-only**: no TypeScript runtime code, just `agent.cordis.yml` (the flatted agent entry list) plus `preset.yml` metadata.
+
+## What it does
+
+- **A full standard-mode baseline.** `agent.cordis.yml` opens with a verbatim copy of the stock `standard` preset (the 16 baseline rows), so a CC-mode session keeps every capability of the standard coding agent.
+- **A scoped CC surface.** The `cc rows` section mounts the Claude Code parity plugins — hook bridge (18 of 30 hook events), CC plugin-directory glue, `ToolSearch`, memory + consolidation, output styles, the coordinator, 21 slash commands, and the worktree/sleep/notebook/structured-output and git tools. WebFetch: stock `tool-web` runs with `fetch: false` and the 60s search timeout carried forward; `web_fetch` is replaced by the cc-services `tool-web-fetch` row (`@dsh-cc/tool-web-fetch`, optional `prompt` summarized on the cheap lane). The fetch executor is the stock dsh-base provider (DNS-pinned, public-only per hop) with CC egress caps restated in the cc-shell patch, and the tool upgrades `http:` URLs to `https:` (CC parity); an optional prompt without the `haiku` route hard-fails before any fetch.
+- **Session titles and `/rename`.** The cc-shell bundle swaps the stock `session-title-llm` host row for `@dsh-cc/session-title-provider`, a first-prompt title provider that stamps the `haiku` cheap lane when configured (else it inherits the main route), and this preset mounts `@dsh-cc/command-rename` so `/rename <title>` pins a user title.
+- **Isolated service realms.** Service-bearing rows (tool search, microcompactor, plugin registry, MCP connections, hook-bridge status, plugin MCP seam) live inside the `cc-services` group with the seven required `isolate` keys, so they publish into an entry-local realm instead of the process-global root realm (which would trip the preset mount gate).
+- **CC Task delegation (`tool-task` + `cc-model-routes`).** The `cc-services` group also mounts `@dsh-cc/subagent-task` (the CC Task tool, `subagent_type` dispatch over per-workspace `.claude/agents` definitions) and `@dsh-cc/model-aliases` (`cc-model-routes`, owning the `model-aliases` settings namespace and the spawn-time alias resolver). In this preset the harness `tool-subagent` and `tool-subagent-fork` rows are **disabled** in favour of the replacement Task, which also owns the durable background/`continuable` loop: `run_in_background: true` starts an agentId-addressable continuable child, and `report`/`send_message`/`list_agents`/`interrupt_agent` operate on that id (`subagent_type: "fork"` stays foreground one-shot until upstream harness issue #2124 resolves). See the task package README and the parity matrix for the known limits.
+
+These rows previously lived in the global `cc-shell` patch (`packages/bundle/cc-shell/cordis.patch.yml`) and leaked into every agent preset. Scoping them to this preset is exactly what isolates each surface.
+
+## Install
+
+```bash
+bash scripts/sync-local-profile.sh web   # mirror @dsh-cc/* packages into the profile
+bash scripts/sync-cc-preset.sh           # install the cc preset into ~/.dsh/.agent-presets/
+```
+
+Both scripts copy, they do not symlink; plugin code and preset files are read at boot, so **restart dsh** after the first install (subsequent file edits are picked up on the next restart too).
+
+## Select
+
+- **Web UI**: pick "CC mode" from the agent-preset selector; or
+- **settings**: `~/.dsh/settings.json` → `"agent-presets": { "default": "cc" }`.
+- **TUI**: `dsh --profile tui` (or `dsh-cc`) boots the terminal surface with this preset as the default.
+
+## Known limits
+
+1. **Standing mount, host-plane singletons.** A preset is mounted once per process under a standing scope. MCP connections, the on-disk CC plugin directories loaded by the glue, the subagent-provider roster, and CC-plugin `settings.json` writes resolve to process-shared host-plane singletons (the same criterion upstream `subagents`/`goals` already follow), so those are not per-session across simultaneously mounted presets.
+2. **Vendored baseline, drift gate.** The standard baseline is vendored. After upgrading dsh, run the drift gate (`pnpm vitest run packages/preset/cc`, or the binary directly) to re-diff it against the new standard preset and fold in upstream changes. On a CI machine with no dsh install present, the gate auto-skips via `it.runIf`.
+3. **Uninstall is deletion.** Remove `~/.dsh/.agent-presets/cc` (or however it was installed); the four built-in modes are unaffected — a user-root preset of the same `id` never overrides the installed system-root entries.
+4. **`DSH_COORDINATOR_MODE=1` breaks this preset's mount.** The coordinator needs an agent `ctx`, and a standing mount has none; that failure is now scoped to this preset's session creation. In the old global-patch era the whole app failed to boot — the blast radius is narrower, but the mode is still unsupported here.
+5. **A settings default to a missing preset errors.** If the default names a preset that does not exist, session creation reports `agent-preset-not-found` (the `details.available` list names the valid ids). Reset by pointing `~/.dsh/settings.json` → `agent-presets.default` back to `standard`.
+
+## Links
+
+- Source composition: `agent.cordis.yml` (baseline + `cc rows`)
+- Metadata: `preset.yml`
+- Origin rows: `packages/bundle/cc-shell/cordis.patch.yml`
+- Composition contract: `tests/composition.spec.ts`

package/README.zh.md

@@ -0,0 +1,44 @@
+# @dsh-cc/preset-cc
+
+[English](README.md) | 中文
+
+DeepSeek Harness(dsh)的 **CC 模式** agent preset:除内置 `standard`、`minimal` 等之外的第 5 个 agent preset。它在完整标准 preset 之上组合出 Claude Code 行为面——Claude hooks、磁盘上的 CC 插件目录、斜杠命令、memory、CC 输出样式、CC 风格 WebFetch。本包**纯配置**:无 TS 运行时代码,只有 `agent.cordis.yml`(平铺的 agent 组合 entry list)加 `preset.yml` 元数据。
+
+## 作用
+
+- **完整标准基底**。`agent.cordis.yml` 以标准 `standard` preset 的逐字拷贝开头(16 个基座行),CC 模式会话因此具备标准编码 agent 的全部能力。
+- **作用域化的 CC 表面**。`cc rows` 段挂载 Claude Code 对齐插件——hook 桥(30 个 hook 事件中的 18 个)、CC 插件目录 glue、`ToolSearch`、memory 与 consolidation、输出样式、coordinator、21 个斜杠命令,以及 worktree/sleep/notebook/structured-output 与 git 工具。WebFetch:stock `tool-web` 以 `fetch: false` 运行,搜索超时沿用 60s;`web_fetch` 由 cc-services 里的 `tool-web-fetch` 行(`@dsh-cc/tool-web-fetch`,可选 `prompt` 在廉价 lane 上摘要)替代。fetch 执行器是 dsh-base 自带的 stock provider(DNS 钉定、逐跳仅公网),CC 出口上限由 cc-shell patch 重述;工具层会把 `http:` URL 升级为 `https:`(CC 对齐);没有 `haiku` 路由时,可选 `prompt` 会在任何 fetch 之前硬失败。
+- **会话标题与 `/rename`**。cc-shell bundle 用 `@dsh-cc/session-title-provider` 替换原生 `session-title-llm` 宿主行——first-prompt 标题提供方,配置了 `haiku` 廉价通道时盖印该路由(否则继承主路由);本 preset 还挂载 `@dsh-cc/command-rename`,`/rename <title>` 即可固定用户标题。
+- **隔离的服务 realm**。承载服务的行(工具搜索、microcompactor、插件注册表、MCP 连接、hook-bridge 状态、插件 MCP seam)放进 `cc-services` 组并带七个必需的 `isolate` 键,发布到 entry-local realm 而非进程全局 root realm(否则会触发 preset 挂载门禁)。
+- **CC Task 委派(`tool-task` + `cc-model-routes`)**。`cc-services` 组还挂载 `@dsh-cc/subagent-task`(CC Task 工具,对按工作区的 `.claude/agents` 定义做 `subagent_type` 派发)与 `@dsh-cc/model-aliases`(`cc-model-routes`,拥有 `model-aliases` settings 命名空间与派发时别名解析器)。本 preset 里 harness 的 `tool-subagent` 与 `tool-subagent-fork` 两行被**禁用**,改用这个替代 Task,它同样拥有 durable 后台/`continuable` 流程:`run_in_background: true` 启动一个可按 agentId 寻址的 continuable 子代理,`report`/`send_message`/`list_agents`/`interrupt_agent` 都作用于该 id(`subagent_type: "fork"` 在上游 harness issue #2124 解决前保持前台一次性)。已知限制见 task 包 README 与 parity matrix。
+
+这些行此前位于全局 `cc-shell` patch(`packages/bundle/cc-shell/cordis.patch.yml`),会泄漏进每个 agent preset;把它们 scope 到本 preset 正是隔离各行为面的手段。
+
+## 安装
+
+```bash
+bash scripts/sync-local-profile.sh web   # 把 @dsh-cc/* 包镜像到 profile
+bash scripts/sync-cc-preset.sh           # 把 cc preset 装进 ~/.dsh/.agent-presets/
+```
+
+两个脚本都是拷贝而非软链;插件代码与 preset 文件在启动时读取,因此首次安装后需**重启 dsh**(之后的文件改动也在下次重启生效)。
+
+## 选用
+
+- **Web UI**:预设选择器里选「CC mode」;或
+- **settings**:`~/.dsh/settings.json` → `"agent-presets": { "default": "cc" }`。
+
+## Known limits(已知限制)
+
+1. **standing 挂载 + host-plane 单例**。preset 在每个进程挂载一次并处于 standing scope。MCP 连接、glue 装载的磁盘 CC 插件目录、subagent provider 名册、CC 插件对 `settings.json` 的写入都落到进程共享的 host-plane 单例(与上游 `subagents`/`goals` 相同的判据),因此不会因同时挂载的多个 preset 而按会话隔离。
+2. **基座 vendored,靠漂移闸**。标准基座是 vendored 的。升级 dsh 后运行漂移闸(`pnpm vitest run packages/preset/cc`,或直跑二进制)重新 diff 新标准 preset 并合入上游改动。CI 机器无 dsh 安装时闸通过 `it.runIf` 自动跳过。
+3. **卸载即删除**。删掉 `~/.dsh/.agent-presets/cc`(或你的安装方式);四个内置模式不受影响——user root 下同 id 的 preset 也盖不掉已安装的 system root 条目。
+4. **`DSH_COORDINATOR_MODE=1` 使本 preset 挂载失败**。coordinator 需要 agent `ctx`,而 standing 挂载点没有;该失败现在只影响本 preset 的会话创建。在旧全局 patch 时代整个 app 起不来——影响面已收窄,但此模式在此处仍不受支持。
+5. **settings 指向不存在的 preset 会报错**。若默认指向不存在的 preset,建会话报 `agent-preset-not-found`(`details.available` 列出可用 id)。把 `~/.dsh/settings.json` 的 `agent-presets.default` 改回 `standard` 即复位。
+
+## 链接
+
+- 源组合:`agent.cordis.yml`(基座 + `cc rows`)
+- 元数据:`preset.yml`
+- 行来源:`packages/bundle/cc-shell/cordis.patch.yml`
+- 组合契约:`tests/composition.spec.ts`

package/agent.cordis.yml

@@ -0,0 +1,664 @@
+# BEGIN dsh-cc header
+# vendored from @deepseek-ai/dsh@0.1.1-rc.2 config/agent-presets/standard/agent.cordis.yml
+# source: config/agent-presets/standard/agent.cordis.yml (deepseek-harness install)
+#
+# This file is the dsh "CC mode" agent preset composition. The baseline below
+# is a verbatim copy of the standard preset; the `cc rows` section appended at
+# the bottom carries the CC surface rows scoped to this preset (they previously
+# lived in the cc-shell global patch, see
+# $R/packages/bundle/cc-shell/cordis.patch.yml).
+#
+# After upgrading dsh upstream, re-diff this baseline against the new standard
+# preset (the drift gate in tests/composition.spec.ts automates this) and fold
+# in any upstream changes.
+# END dsh-cc header
+# The `standard` agent preset: the full coding agent, mounted once per process.
+#
+# This file is an AGENT-PLANE composition. The roster mounts it ONCE under a
+# standing scope; every session naming it joins by scope parentage, so the
+# tools and prompt sections registered here cover each joined agent while a
+# session's own state stays keyed per Session/Agent inside the plugins. The
+# host composition (`base.cordis.yml` + `web.cordis.yml`) keeps everything a
+# preset must not own: the registries themselves, the sandbox and approval
+# stack, persistence, and the model route.
+#
+# A service row here MUST sit inside a group carrying an `isolate` realm.
+# Without one it publishes into the root realm, where it is process-global —
+# another preset publishing the same name collides, and a host reader would
+# resolve one preset's instance for every session; `dsh-agent-presets` rejects
+# that at mount. `true` means an entry-local realm: this standing mount's own
+# private instance, apart from every other preset's. (A shared label does NOT
+# pool instances — `provide()` throws on the second registration under the
+# same realm symbol; labels join REALMS, and are not what this file needs.)
+
+# ── identity ────────────────────────────────────────────────────────────────
+
+# The preset's own persona, shadowing the deployment default for this agent.
+# `{{model}}` and `{{cwd}}` resolve from the agent's own route and workspace.
+- id: persona
+  name: '@deepseek-ai/dsh-persona'
+  config:
+    prefix: >-
+      You are a coding agent powered by the {{model}} model. Your working directory is {{cwd}}.
+
+- id: agent-instructions
+  name: '@deepseek-ai/dsh-agent-instructions'
+  config:
+    maxBytes: 65536
+
+# ── shell ───────────────────────────────────────────────────────────────────
+
+# `shell-env` stays in the HOST composition: `apps/cli/src/web.ts` injects it to
+# publish `DSH_WEB_URL`/`DSH_WEB_MODE`, and a host row that injects a service is
+# the criterion for host-plane ownership — injection resolves before any session
+# exists, so there is no agent to key by. Behind a preset realm those variables
+# never reached the model's shell at all. Both shell tools consume the host
+# registry from here; their executors (`bash-sandbox`/`pwsh-sandbox`) are
+# host-plane too.
+- id: tool-bash
+  name: '@deepseek-ai/dsh-tool-bash'
+  disabled: !!js process.platform === 'win32'
+
+- id: tool-pwsh
+  name: '@deepseek-ai/dsh-tool-pwsh'
+  disabled: !!js process.platform !== 'win32'
+
+# ── filesystem ──────────────────────────────────────────────────────────────
+
+# Both register into the host `tools` registry and provide nothing, so
+# they need no realm. The `fs` service and its policy stay in the host.
+- id: tool-fs
+  name: '@deepseek-ai/dsh-tool-fs'
+
+- id: tool-fs-search
+  name: '@deepseek-ai/dsh-tool-fs-search'
+  config:
+    sampleOverCapGlobResults: false
+
+# ── background jobs ────────────────────────────────────────────────────────
+
+# Only the model-facing controls. The task REGISTRY stays on the host plane:
+# its producers sit outside any realm this file could put it in — `tool-bash`
+# above resolves it with `ctx.get`, and an entry-local realm here is invisible
+# to every sibling row, so `run_in_background` would answer "background jobs
+# unavailable" while these controls sat in the catalog. The registry is keyed by
+# owning agent anyway, so one host instance serves every session. What a preset
+# chooses is whether its agent can collect and stop background work at all.
+- id: tool-jobs
+  name: '@deepseek-ai/dsh-tool-jobs'
+
+# ── skills ──────────────────────────────────────────────────────────────────
+
+# The skill REGISTRY lives in the host composition and is layered per scope:
+# these rows register into THIS preset's layer of it, so they need no realm.
+# `skill-filesystem` contributes local-root discovery for agents on this preset, and
+# `tool-skill` gives them the catalog and loader; the merged catalog also
+# carries whatever the deployment registered globally (repository plugins).
+- id: skill-filesystem
+  name: '@deepseek-ai/dsh-skill-filesystem'
+
+- id: tool-skill
+  name: '@deepseek-ai/dsh-tool-skill'
+
+# ── goals ───────────────────────────────────────────────────────────────────
+
+# Only the model-facing tool. The goal SERVICE, its session driver, and the
+# `/goal` command stay on the host plane: the Gateway serves the goal domain as
+# Remote endpoints whose receiver comes from a generated descriptor, so it
+# resolves `goals` on the host and an entry-local realm here would hide it. The
+# registry is keyed by session anyway, so one host instance serves every
+# session. What a preset chooses is whether its agent can call the goal tool.
+- id: tool-goal
+  name: '@deepseek-ai/dsh-tool-goal'
+
+# ── plan mode ───────────────────────────────────────────────────────────────
+
+# Plan state is per-agent by nature, so an entry-local realm is not a
+# workaround here — it is the correct lifetime.
+- id: planning
+  name: cordis:group
+  group: true
+  isolate:
+    planMode: true
+  config:
+    - id: plan-mode
+      name: '@deepseek-ai/dsh-plan-mode'
+      config:
+        section: |
+              You are in plan mode. Stay in plan mode until exit_plan_mode succeeds or the user switches the session mode. Imperative language to implement changes means plan the implementation, not execute it. A user's conversational agreement — including an answer confirming something you asked — approves nothing and does not end plan mode; fold the confirmed decision into the plan and submit it through exit_plan_mode.
+
+              Explore first. Use non-mutating reads, searches, static analysis, and checks to ground the plan in the actual repository. Do not edit or write files, change configuration, run formatters or code generation that rewrites tracked files, commit, or otherwise carry out the plan. Prefer existing functions and patterns over new machinery.
+
+              The tool catalog stays the same across modes for request-cache stability. These plan-mode rules override any later tool description or guidance that suggests using mutation tools; those tools remain listed to keep the tool catalog unchanged. Do not use todo_write to track this planning phase: it tracks implementation after an approved plan, while the plan itself belongs in exit_plan_mode.
+
+              Resolve discoverable facts by inspection. Use ask_user_question only for user-owned choices or material ambiguity that inspection cannot answer. Do not ask the user where code lives or how current behavior works when you can find out.
+
+              Make the plan decision-complete: state the goal and success criteria; group implementation changes by subsystem; identify public API, schema, and data-flow changes; cover edge cases, failure modes, tests, acceptance criteria, and explicit assumptions. Keep it concise enough to review but detailed enough that another engineer can implement it without making design decisions.
+
+              When ready, call exit_plan_mode with the complete plan markdown, starting with a # title. Make exit_plan_mode the only and final tool call in that assistant response: it presents the plan for approval, and implementation begins only in a later step after approval. Do not paste the final plan as a plain reply or ask "should I proceed?" through prose or ask_user_question. If review rejects it, incorporate the feedback and present again. If the review channel is unavailable or aborted, stay in plan mode and ask the user to switch modes manually; do not proceed with implementation.
+
+# ── compaction ──────────────────────────────────────────────────────────────
+
+# `compaction-basic` reads `toolResultPrune` through `ctx.get`, so the pruner must
+# share this realm rather than sit outside it.
+#
+# The engine and the command are CC replacements for the upstream rows: the
+# engine subclass folds a per-agent `/compact [instructions]` hint into the
+# summarizer input; the command forwards the free-text argument as that hint
+# instead of rejecting arguments.
+#
+# `tokenMeter` is deliberately NOT in this realm: the meter stays on the HOST
+# plane, and the rows here resolve that one instance. It takes no configuration,
+# keys every fold by Session, and owns the context-meter projection units the
+# browser reads for every session — behind a realm those units would come and go
+# with whichever presets happen to be mounted. What a preset chooses is whether
+# its agent compacts at all, which is `compaction-basic` below.
+- id: compaction
+  name: cordis:group
+  group: true
+  isolate:
+    compaction: true
+    toolResultPruner: true
+  config:
+    - id: compaction-basic
+      name: '@dsh-cc/compaction-basic'
+
+    - id: command-compact
+      name: '@dsh-cc/command-compact'
+
+    - id: tool-result-pruner
+      name: '@deepseek-ai/dsh-compaction-tool-result-pruner'
+      config:
+        thresholdChars: 8192
+        headChars: 4096
+        tailChars: 1024
+
+# ── delegation and workflows ────────────────────────────────────────────────
+
+# The `subagents` registry and its spawn/fork backends live in the HOST
+# composition: the registry is a process singleton whose cross-session queries
+# the api-proxy serves to the browser, and a provider name may only be
+# registered once. This preset contributes the delegation TOOLS, which resolve
+# that host registry.
+#
+# `workflows` is different — nothing outside an agent reads it — so every row
+# that reaches it shares one entry-local realm here, and a consumer left
+# outside would resolve a host registry this preset does not populate.
+#
+# `tool-subagent-report` is GONE upstream: there is no report tool anymore.
+# Workers return outcomes to their parent with the host-plane
+# `tool-subagent-control` send_message tool (addressed to the parent's session
+# id), and settlement reaches the parent as the subagent manager's
+# `subagent-settled` notice. `tool-subagent-control` stays host-plane for the
+# same reason as the registry: its tools resolve `ctx.subagents.sendMessage` /
+# interrupt state on the host plane, not a tool this agent registers locally.
+#
+# `tool-subagent` and `tool-subagent-fork` (the harness spawn/fork Task tools)
+# are DISABLED here — the CC Task tool is provided by the `tool-task` row in
+# the cc-services group instead. That replacement Task now implements the
+# continuable loop itself: `run_in_background: true` starts a durable
+# background child on the spawn provider, and send_message /
+# list_agents / interrupt_agent operate on its returned agentId. The harness
+# rows stay disabled. Note for future readers: the disabled
+# `tool-subagent-fork` row's `backgroundMode: continuable` config is inert
+# while the row is disabled, and fork children remain foreground one-shot
+# until upstream harness issue #2124 resolves.
+- id: delegation
+  name: cordis:group
+  group: true
+  isolate:
+    workflowEngine: true
+  config:
+    - id: tool-subagent-control
+      name: '@deepseek-ai/dsh-tool-subagent-control'
+
+    - id: tool-subagent-list-agents
+      name: '@deepseek-ai/dsh-tool-subagent-control/list-agents'
+
+    - id: tool-subagent
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: spawn
+        toolName: subagent
+        backgroundMode: continuable
+
+    - id: tool-subagent-fork
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: fork
+        toolName: subagent_fork
+        backgroundMode: continuable
+
+    # Production dsh does not install these optional providers. Install the
+    # matching Bundle in this Profile and restart the Host, then copy this
+    # preset and remove `disabled` from the matching tool row. Host availability
+    # alone grants no tool.
+    - id: tool-subagent-codex
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: codex
+        toolName: subagent_codex
+        backgroundMode: one-shot
+        maxDepth: provider-managed
+
+    - id: tool-subagent-claude-code
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: claude-code
+        toolName: subagent_claude_code
+        backgroundMode: one-shot
+        maxDepth: provider-managed
+
+    - id: workflow-worker-thread
+      name: '@deepseek-ai/dsh-workflow-worker-thread'
+      config:
+        provider: spawn
+
+    - id: tool-workflow
+      name: '@deepseek-ai/dsh-tool-workflow'
+
+    - id: tool-ralph
+      name: '@deepseek-ai/dsh-tool-ralph'
+      config:
+        subagentProvider: spawn
+        maxRounds: 64
+
+# ── remaining model-facing rows ─────────────────────────────────────────────
+
+- id: tool-ask-user
+  name: '@deepseek-ai/dsh-tool-ask-user'
+
+- id: tool-todo
+  name: '@deepseek-ai/dsh-tool-todo'
+  config:
+    allowParallelInProgress: true
+
+# The `web` service and its search provider stay in the host composition; only
+# the model-facing tool is per-session.
+#
+# Moved in from the cc-shell global swap. fetch is still disabled here because
+# `@dsh-cc/tool-web-fetch` inside cc-services owns the model-facing tool
+# (adds the optional `prompt` and upgrades http: URLs to https: before the
+# seam). The HTTP executor is the stock `@deepseek-ai/dsh-web-fetch-http`
+# provider from dsh-base, with CC egress caps restated in cc-shell's patch.
+# A prompt without the `haiku` route hard-fails.
+- id: tool-web
+  name: '@deepseek-ai/dsh-tool-web'
+  config:
+    fetch: false
+    searchTimeoutMs: 60000
+
+# ── cc rows ─────────────────────────────────────────────────────────────────
+
+# The CC surface rows, scoped to this preset. They previously lived in the
+# cc-shell global patch ($R/packages/bundle/cc-shell/cordis.patch.yml) and are
+# moved here so they no longer leak into every agent preset that mounts the
+# global host composition.
+
+# Service rows MUST sit in a group with an `isolate` realm: mounting this preset
+# rejects any row that publishes a cordis Service into the root realm
+# (PresetMountError / leakedServices). The isolate keys below are all required —
+# the glue dynamically attaches mcp-client as a sub-fiber, and `Context.extend`
+# inherits the group's isolate map down the prototype chain, so without
+# `mcpConnections` isolated here it would publish into the root realm and break
+# the preset. The command rows and the `memory` / `hooks-claude-code` consumers
+# must live INSIDE the group: they `ctx.get('ccModelRoutes')` (and
+# `ctx.get('mcpConnections')`), which resolve to nothing outside the isolated
+# realm. `ccPlugins` is deliberately NOT isolated: the glue publishes it into
+# the ROOT realm from the root fiber (see CcPluginsService in
+# @dsh-cc/bundle-shell) so host-plane siblings — the TUI driver and /help
+# — can resolve it; isolating the name here would shadow that publication and
+# break in-group resolution too. `mcp` (the plugin MCP seam) IS isolated:
+# only in-group loader fibers `ctx.get('mcp')`; a root-realm provide would
+# trip leakedServices (the providing fiber is the cc-mcp-seam child, inside
+# the preset mount). `hooks` (the plugin hooks seam) IS isolated like `mcp`:
+# only in-group loader fibers ctx.get('hooks'); the providing fiber is
+# hooks-claude-code, inside this group — a root-realm provide would trip
+# leakedServices.
+# The bridge's second provide, `hookRun` (WorktreeCreate / WorktreeRemove
+# invocation; docs/plans/2026-09-14-cc-worktree-parity.md WS-6), is
+# deliberately NOT isolated, mirroring `ccPlugins`: its consumer
+# (tool-git-worktree) mounts top-level, outside this group's realm — and like
+# ccPlugins it is provided from the ROOT fiber (see the bridge's root-provide
+# in @dsh-cc/hooks-claude-code), because a plain in-realm provide here would
+# publish a preset fiber's service into the root realm and trip the preset
+# leakedServices gate (agent-preset/invalid at mount).
+- id: cc-services
+  name: cordis:group
+  group: true
+  isolate:
+    toolSearch: true
+    microcompactor: true
+    ccModelRoutes: true
+    resumePinStore: true
+    mcpConnections: true
+    hookBridgeStatus: true
+    mcp: true
+    hooks: true
+    contextCrusher: true
+  config:
+    # ToolSearch: model-called search that activates deferred tools per scope.
+    - id: tool-search
+      name: '@dsh-cc/tool-search'
+
+    # Model-free staleness microcompaction (auto: false by default).
+    - id: compaction-micro
+      name: '@dsh-cc/compaction-micro'
+
+    # CCR tool-output compression (plan docs/plans/2026-09-10-ccr-tool-output-compression.md):
+    # a `tools/post-execute` listener registered with `prepend: true`, composed
+    # post-`next()`, plus the agent-visible `context_retrieve` tool (listener +
+    # tool host, no consumed service of its own — but it mounts a Service, so it
+    # shares this realm via the `contextCrusher` isolate key above). MUST sit
+    # BEFORE the hooks-claude-code row: waterfall listeners dispatch
+    # outermost-first in registration order, and the crusher must be outermost
+    # so its post-`next()` compression wraps the bridge's PostToolUse fold (the
+    # bridge sees the original bytes; the committed result carries the
+    # compressed form). The order tripwire is pinned by
+    # packages/context/context-crusher/tests/composition.spec.ts.
+    - id: context-crusher
+      name: '@dsh-cc/context-crusher'
+
+    # The hook bridge: runs the Claude Code hooks.json / settings hook config
+    # on the harness interception seams and provides the `hooks` seam plugin
+    # hooks merge through. Must mount BEFORE cc-shell-glue: the glue's
+    # CcPluginsService.mountAll probes ctx.get('hooks') at boot, so the bridge
+    # must have provided it first. Nothing between glue and bridge consumes
+    # bridge services at mount time (command-doctor reads hookBridgeStatus
+    # lazily at invocation), and the bridge's interception listeners use
+    # prepend:true, so ordering vs permission-rules is compose-order-insensitive.
+    - id: hooks-claude-code
+      name: '@dsh-cc/hooks-claude-code'
+      # Tracked root-level hook config (serena code-intelligence plan Phase 0).
+      # Relative configPath resolves against the process launch cwd, so a
+      # worktree session started at the worktree root loads it by construction.
+      config:
+        configPath: hooks.json
+
+    # CC plugins on disk (agents/skills/commands/hooks/mcp servers) plus the
+    # base agent preset dirs and .mcp.json server wiring — see the glue plugin.
+    - id: cc-shell-glue
+      name: '@dsh-cc/bundle-shell'
+
+    - id: command-plugin
+      name: '@dsh-cc/command-plugin'
+
+    - id: command-mcp
+      name: '@dsh-cc/command-mcp'
+
+    # Serena-first prompt steering: while a serena MCP server is ready
+    # (registry-only, live detection over the isolated `mcpConnections`
+    # above), the registered `serena-first` section (order 105) renders the
+    # policy paragraph and the assemble waterfall rewrites the upstream
+    # `tool:read` / `tool:grep` texts (replace-not-mutate, `{ prepend: true }`).
+    # Publishes no Service, so it needs no isolate key of its own;
+    # `systemPrompt` resolves through the group via the prototype chain.
+    # `serverName` config (default 'serena') drives detection and every
+    # emitted `mcp__<serverName>__*` name.
+    - id: serena-first
+      name: '@dsh-cc/serena-first'
+
+    # /doctor reads ccModelRoutes, mcpConnections, and hookBridgeStatus, so it
+    # shares the group realm.
+    - id: command-doctor
+      name: '@dsh-cc/command-doctor'
+
+    # cc-model-routes owns the model-aliases settings namespace and resolver;
+    # tool-task provides the CC Task tool (subagent_type dispatch over
+    # per-workspace .claude/agents) and reserves the names 'subagent'/'workflow'.
+    - id: cc-model-routes
+      name: '@dsh-cc/model-aliases'
+
+    # Resume pins (plan docs/plans/2026-09-04-subagent-resume-pins.md §4.10):
+    # one PinStore + the resume gate/overlay/policy per mounted preset, exposed
+    # as the `resumePinStore` service (isolated above, like ccModelRoutes). The
+    # pins root COLOCATES with the session-persistence root: the harness base
+    # patch mounts the jsonl backend at `!!js dshHomePath('sessions')`, and this
+    # `!!js` expression resolves `dshHomePath` the same way at entry activation
+    # (app-boot provides it on the root ctx before the config tree mounts), so
+    # pins always sit next to the durable sessions they pin. Must mount BEFORE
+    # tool-task: the Task plugin's spawn capture prefers this service store.
+    - id: cc-resume-pins
+      name: '@dsh-cc/subagent-resume-pins'
+      config:
+        pinsRoot: !!js dshHomePath('sessions', 'resume-pins')
+
+    - id: tool-task
+      name: '@dsh-cc/subagent-task'
+
+    # /agents: query/control surface for continuable background children
+    # (plan docs/plans/2026-09-05-continuable-background-ux.md §3.2). Mounts
+    # INSIDE cc-services (F8): the read-only snapshot joins the realm-interior
+    # resumePinStore with the host-plane subagents/agents registries, then
+    # republishes the snapshot as the ROOT-realm `ccAgents` service (CcPlugins
+    # pattern) so the TUI local-slash path consumes the SAME snapshot. Must
+    # mount AFTER cc-resume-pins.
+    - id: command-agents
+      name: '@dsh-cc/command-agents'
+
+    # Memory and the hook bridge consume `ctx.get('ccModelRoutes')` at spawn
+    # time (recall small-fast lane / hook model resolution), so they share this
+    # realm. memory-consolidation stays OUTSIDE: its forks must keep inheriting
+    # the parent route. memory still publishes no cordis Service;
+    # hooks-claude-code now provides `hookBridgeStatus` (for /doctor), which is
+    # isolated on this group's isolate map — both stay mount-legal under the
+    # leakedServices gate only because they sit INSIDE the group.
+    - id: memory
+      name: '@dsh-cc/memory'
+
+    # Replaces the stock dsh-tool-web web_fetch (fetch: false above): adds the
+    # optional `prompt`, summarized on the cheap lane when the haiku alias is
+    # configured; degrades to raw converted text + notice otherwise.
+    - id: tool-web-fetch
+      name: '@dsh-cc/tool-web-fetch'
+
+    # Subagent handoff store (plan docs/plans/2026-09-10-subagent-handoff-store.md):
+    # the handoff_put / handoff_get tool pair over a content-addressed store
+    # under $DSH_HOME/handoff/<projectKey>/. Publishes no Service (plain plugin,
+    # memory pattern), so it needs no isolate key of its own; state is entirely
+    # on disk (TTL 24h + 500-entry disk-based LRU swept on put), so spawned
+    # children in separate processes see the same store.
+    - id: handoff-store
+      name: '@dsh-cc/handoff-store'
+
+    # Stage-0 reasoning-fold probe (plan docs/plans/2026-09-10-reasoning-fold-deepseek.md):
+    # a READ-ONLY `llm/stream` listener byte-counting reasoning vs text deltas
+    # into $DSH_HOME/reasoning-fold/<sessionId>.jsonl. Publishes no Service
+    # (plain plugin, handoff-store pattern), so it needs no isolate key of its
+    # own; the listener never rewrites the chunk stream (LEDGER-ONLY).
+    - id: reasoning-fold
+      name: '@dsh-cc/reasoning-fold'
+
+# ── parallel rows ──────────────────────────────────────────────────────────
+
+# Claude Code skills from user/project/additional roots.
+- id: skill-claude-code
+  name: '@dsh-cc/skill-loader'
+
+- id: memory-consolidation
+  name: '@dsh-cc/memory-consolidation'
+
+# Render CLAUDE.md-style output styles to the system prompt.
+- id: cc-output-styles
+  name: '@dsh-cc/output-styles'
+
+# Coordinator mode: DSH_COORDINATOR_MODE=1 or config enabled.
+- id: coordinator
+  name: '@dsh-cc/coordinator'
+
+- id: schedule
+  name: '@deepseek-ai/dsh-schedule'
+
+- id: session-cwd
+  name: '@dsh-cc/session-cwd'
+
+- id: tool-git-worktree
+  name: '@dsh-cc/tool-git-worktree'
+
+- id: tool-sleep
+  name: '@dsh-cc/tool-sleep'
+
+# NotebookEdit: edits Jupyter notebook (.ipynb) cells over the ctx.fs seam
+# with CC's replace/insert/delete, cell addressing, and read-before-write gate.
+- id: tool-notebook-edit
+  name: '@dsh-cc/tool-notebook-edit'
+
+# StructuredOutput: non-interactive structured-final-response channel, registered
+# only when the deployment declares a schema (CC's SyntheticOutputTool).
+- id: tool-structured-output
+  name: '@dsh-cc/tool-structured-output'
+
+# ── CC parity slash commands ───────────────────────────────────────────────
+
+# CacheHealth: passive prompt-cache prefix-stability observer; /cache-health
+# joins its own llm/stream observation ledger with provider-metered cache
+# usage from session events (detector-only, no request rewriting).
+- id: cache-health
+  name: '@dsh-cc/cache-health'
+
+- id: command-cost
+  name: '@dsh-cc/command-cost'
+  config:
+    # Composition decides pricing → the package schema stays strict. The preset
+    # ships a starter table of official published list prices (USD per 1M
+    # tokens), collected 2026-09-10 from the vendor pricing pages; deployments
+    # whose billing differs should override the whole modelTable. No '*'
+    # wildcard column on purpose: unmatched models report "no price
+    # configured" instead of a misleading zero cost. A runtime id with a route
+    # prefix (e.g. `llmbox_ant/glm-5.3`) matches the bare model row via suffix
+    # matching.
+    modelTable:
+      # DeepSeek (api-docs.deepseek.com/quick_start/pricing); cache write is
+      # not billed separately → cacheWrite 0.
+      - model: deepseek-v4-flash-0731
+        inputPerMTok: 0.28
+        outputPerMTok: 0.42
+        cacheReadPerMTok: 0.04
+        cacheWritePerMTok: 0
+      - model: deepseek-v4-flash
+        inputPerMTok: 0.28
+        outputPerMTok: 0.42
+        cacheReadPerMTok: 0.04
+        cacheWritePerMTok: 0
+      - model: deepseek-v4-pro
+        inputPerMTok: 0.56
+        outputPerMTok: 1.68
+        cacheReadPerMTok: 0.08
+        cacheWritePerMTok: 0
+      # GLM (docs.z.ai/guides/overview/pricing); cached-input storage is
+      # limited-time free → cacheWrite 0. glm-5.3-flash uses list prices, not
+      # the 50%-off promo.
+      - model: glm-5.3
+        inputPerMTok: 1.4
+        outputPerMTok: 4.4
+        cacheReadPerMTok: 0.26
+        cacheWritePerMTok: 0
+      - model: glm-5.2
+        inputPerMTok: 1.4
+        outputPerMTok: 4.4
+        cacheReadPerMTok: 0.26
+        cacheWritePerMTok: 0
+      - model: glm-5.1
+        inputPerMTok: 1.4
+        outputPerMTok: 4.4
+        cacheReadPerMTok: 0.26
+        cacheWritePerMTok: 0
+      - model: glm-5
+        inputPerMTok: 1.0
+        outputPerMTok: 3.2
+        cacheReadPerMTok: 0.2
+        cacheWritePerMTok: 0
+      - model: glm-4.7
+        inputPerMTok: 0.6
+        outputPerMTok: 2.2
+        cacheReadPerMTok: 0.11
+        cacheWritePerMTok: 0
+      - model: glm-4.6
+        inputPerMTok: 0.6
+        outputPerMTok: 2.2
+        cacheReadPerMTok: 0.11
+        cacheWritePerMTok: 0
+      - model: glm-4.5
+        inputPerMTok: 0.6
+        outputPerMTok: 2.2
+        cacheReadPerMTok: 0.11
+        cacheWritePerMTok: 0
+      - model: glm-5.3-flash
+        inputPerMTok: 0.15
+        outputPerMTok: 0.50
+        cacheReadPerMTok: 0.03
+        cacheWritePerMTok: 0
+      # Kimi (kimi.ai kimi-k3 pricing page).
+      - model: kimi-k3
+        inputPerMTok: 3.00
+        outputPerMTok: 15.00
+        cacheReadPerMTok: 0.30
+        cacheWritePerMTok: 0
+
+- id: command-export
+  name: '@dsh-cc/command-export'
+
+- id: command-stats
+  name: '@dsh-cc/command-stats'
+
+- id: command-status
+  name: '@dsh-cc/command-status'
+
+- id: command-memory
+  name: '@dsh-cc/command-memory'
+
+- id: command-skills
+  name: '@dsh-cc/command-skills'
+
+- id: command-help
+  name: '@dsh-cc/command-help'
+
+- id: command-config
+  name: '@dsh-cc/command-config'
+
+- id: command-permissions
+  name: '@dsh-cc/command-permissions'
+
+- id: command-version
+  name: '@dsh-cc/command-version'
+
+- id: command-release-notes
+  name: '@dsh-cc/command-release-notes'
+
+- id: command-diff
+  name: '@dsh-cc/command-diff'
+
+- id: command-init
+  name: '@dsh-cc/command-init'
+
+- id: command-learn
+  name: '@dsh-cc/command-learn'
+
+- id: command-tasks
+  name: '@dsh-cc/command-tasks'
+
+- id: command-resume
+  name: '@dsh-cc/command-resume'
+
+- id: command-rename
+  name: '@dsh-cc/command-rename'
+
+- id: command-branch
+  name: '@dsh-cc/command-branch'
+
+# Cache-stable tool ordering: remembers each scope's last emitted tool-name
+# sequence on the system-prompt/assemble waterfall and appends newly appeared
+# tools (ToolSearch activations, MCP registrations) lexicographically at the
+# tail, so an activation extends the request prefix instead of shifting every
+# tool after the insertion point. cordis waterfall is outermost-first and the
+# outermost listener has the final say, so the plugin registers its assemble
+# listener with `prepend` to stay outermost regardless of row position; the
+# row itself stays last because the composition drift gate expects cc rows at
+# the bottom (the Item 6 L0 prefix-stability e2e is the sentinel).
+# Publishes no Service, so it needs no isolate realm.
+- id: tool-append-order
+  name: '@dsh-cc/tool-append-order'

package/package.json

@@ -0,0 +1,76 @@
+{
+  "name": "@dsh-cc/preset-cc",
+  "description": "dsh CC mode agent preset: composition-only package mounting the Claude Code surface",
+  "version": "0.7.1-rc.1",
+  "type": "module",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/dsh-cc/dsh-cc.git",
+    "directory": "packages/preset/cc"
+  },
+  "files": [
+    "agent.cordis.yml",
+    "preset.yml",
+    "README.md",
+    "README.zh.md"
+  ],
+  "license": "Apache-2.0",
+  "publishConfig": {
+    "access": "public"
+  },
+  "dependencies": {
+    "@dsh-cc/command-agents": "^0.7.1-rc.1",
+    "@dsh-cc/command-compact": "^0.7.1-rc.1",
+    "@dsh-cc/command-branch": "^0.7.1-rc.1",
+    "@dsh-cc/command-config": "^0.7.1-rc.1",
+    "@dsh-cc/command-cost": "^0.7.1-rc.1",
+    "@dsh-cc/bundle-shell": "^0.7.1-rc.1",
+    "@dsh-cc/cache-health": "^0.7.1-rc.1",
+    "@dsh-cc/command-diff": "^0.7.1-rc.1",
+    "@dsh-cc/command-help": "^0.7.1-rc.1",
+    "@dsh-cc/command-export": "^0.7.1-rc.1",
+    "@dsh-cc/command-init": "^0.7.1-rc.1",
+    "@dsh-cc/command-doctor": "^0.7.1-rc.1",
+    "@dsh-cc/command-learn": "^0.7.1-rc.1",
+    "@dsh-cc/command-mcp": "^0.7.1-rc.1",
+    "@dsh-cc/command-permissions": "^0.7.1-rc.1",
+    "@dsh-cc/command-memory": "^0.7.1-rc.1",
+    "@dsh-cc/command-plugin": "^0.7.1-rc.1",
+    "@dsh-cc/command-release-notes": "^0.7.1-rc.1",
+    "@dsh-cc/command-resume": "^0.7.1-rc.1",
+    "@dsh-cc/command-skills": "^0.7.1-rc.1",
+    "@dsh-cc/command-rename": "^0.7.1-rc.1",
+    "@dsh-cc/command-stats": "^0.7.1-rc.1",
+    "@dsh-cc/command-status": "^0.7.1-rc.1",
+    "@dsh-cc/command-tasks": "^0.7.1-rc.1",
+    "@dsh-cc/compaction-basic": "^0.7.1-rc.1",
+    "@dsh-cc/command-version": "^0.7.1-rc.1",
+    "@dsh-cc/compaction-micro": "^0.7.1-rc.1",
+    "@dsh-cc/context-crusher": "^0.7.1-rc.1",
+    "@dsh-cc/coordinator": "^0.7.1-rc.1",
+    "@dsh-cc/memory": "^0.7.1-rc.1",
+    "@dsh-cc/memory-consolidation": "^0.7.1-rc.1",
+    "@dsh-cc/hooks-claude-code": "^0.7.1-rc.1",
+    "@dsh-cc/model-aliases": "^0.7.1-rc.1",
+    "@dsh-cc/session-cwd": "^0.7.1-rc.1",
+    "@dsh-cc/output-styles": "^0.7.1-rc.1",
+    "@dsh-cc/skill-loader": "^0.7.1-rc.1",
+    "@dsh-cc/serena-first": "^0.7.1-rc.1",
+    "@dsh-cc/subagent-resume-pins": "^0.7.1-rc.1",
+    "@dsh-cc/subagent-task": "^0.7.1-rc.1",
+    "@dsh-cc/tool-git-worktree": "^0.7.1-rc.1",
+    "@dsh-cc/tool-append-order": "^0.7.1-rc.1",
+    "@dsh-cc/tool-sleep": "^0.7.1-rc.1",
+    "@dsh-cc/tool-notebook-edit": "^0.7.1-rc.1",
+    "@dsh-cc/tool-structured-output": "^0.7.1-rc.1",
+    "@dsh-cc/tool-search": "^0.7.1-rc.1",
+    "@dsh-cc/handoff-store": "^0.7.1-rc.1",
+    "@dsh-cc/tool-web-fetch": "^0.7.1-rc.1",
+    "@dsh-cc/reasoning-fold": "^0.7.1-rc.1"
+  },
+  "devDependencies": {
+    "@deepseek-ai/cordis-plugin-include": "link:../../../../deepseek-harness/vendor/include",
+    "js-yaml": "^4.1.0",
+    "vitest": "^4.1.8"
+  }
+}

package/preset.yml

@@ -0,0 +1,3 @@
+name: CC mode
+description: Full Standard-mode capability, plus the Claude Code surface — .claude plugin directories, Claude hooks, slash commands, persistent memory, output styles, and fetch-enabled Web tools.
+order: 5