Merge origin/master into feat/mcp-auto-reconnect

Integrate master at 8b4ef532 before addressing review feedback so each bot issue remains isolated in its own descendant commit. Resolve the vendor package rescope by moving the PR's new MCP supervisor imports to @deepseek-ai/cordis and @deepseek-ai/schemastery, preserve the reconnect implementation and e2e coverage, and regenerate the merged config and module catalogs with fresh bilingual pairing records.
This commit is contained in:
Tianyi Cui
2026-08-10 23:44:58 +08:00
1749 changed files with 12061 additions and 9269 deletions

View File

@@ -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 .agents/notes/implemented/feature/2026-07-06-approval-seam.md
2026-07-06-approval-seam.md: 7c830d93f19a40ab193cfebabca854882ab68d62
2026-07-06-approval-seam.zh.md: 9dedfddadc23b0da44b28e8750508653ee20bb83
2026-07-06-approval-seam.md: 8aa9986139dae77e08c166b72545bfa688a389e0
2026-07-06-approval-seam.zh.md: ef4ccf5fd2b54888a648737866ff6f5fe1678882

View File

@@ -123,7 +123,7 @@ Costs and accepted limits:
- **Who decides whether a call asks in the first place?** Policy producers: a hook returning `permissionDecision: ask`, any `tools/pre-execute` listener, or the sandbox escalation gate. The seam and the bridge only route and answer; neither injects its own judgment about what deserves a prompt.
- **What happens when the user dismisses the prompt, or the turn aborts mid-ask?** Dismissal maps to `cancelled` with its own deny text. An already-aborted signal settles `cancelled` without dispatching; an abort during the ask discards the late answer. When both audit appends commit, either path records one pair, never two.
- **What if the client answers with an option the harness never offered?** Any selection other than the offered `allow_once` maps to `rejected` — an unknown optionId from a non-conforming client can never grant.
- **How do subagents' approvals route?** An agent no answerer owns delegates through the whole waterfall and fails closed — in-process subagents are deliberately unanswerable. A `'never'` parent seeds that override into each in-process child's log ([decision](2026-07-25-subagent-policy-inheritance.md)), so the child is told up front instead of asking into the empty waterfall. `subagent-acp`'s child-side auto-answer is separate; routing a child's asks to the parent controller is deferred (§ Deferred).
- **How do subagents' approvals route?** They do not: delegation pins every in-process child to `'never'` ([approvals-pinned decision](2026-08-10-subagent-approval-pinned-never.md)), so each child ask resolves `rejected` before any answerer and the child is told up front through its runtime context. `subagent-acp`'s child-side auto-answer is separate; routing a child's asks to the parent controller is deferred (§ Deferred).
- **What does `policy: 'never'` actually change at runtime?** The service resolves every ask for that session to `rejected` before dispatching any answerer (in-service, so no registration order can bypass it); the next atomic runtime-context snapshot states the policy; each successful auto-rejection records the audit pair.
- **What happens across a hot reload, or when an answerer unloads mid-session?** Answerers dispose with their owning fiber, so the next ask degrades to `unavailable` instead of hanging on a dead channel; remounting re-registers the answerer with no catch-up state.
- **Where does a client get approval context?** The request carries the exact `callId` and the asker's human-readable `reason`; channel adapters may correlate richer tool-call state without duplicating arguments in the approval seam.

View File

@@ -123,7 +123,7 @@ ACP 桥只应答其会话映射所拥有的精确 agent 对象。它携带既有
- **谁决定一次调用是否需要 ask?** 策略生产者:返回 `permissionDecision: ask` 的钩子、任何 `tools/pre-execute` 监听器、或沙箱升级门禁。seam 和桥只负责路由和应答;二者都不注入自己对「什么值得弹出提示」的判断。
- **用户关闭提示或轮次在 ask 进行中中止时会发生什么?** 关闭映射为 `cancelled` 并携带自己的拒绝文本。已中止的 signal 直接结算为 `cancelled` 而不派发;ask 进行中的中止丢弃迟到的应答。当两个审计追加都提交时,任一路径都记录恰好一对事件,绝不会两对。
- **如果客户端以 harness 从未提供的选项应答呢?** 除已提供的 `allow_once` 之外的任何选项都映射为 `rejected`——来自不合规客户端的未知 optionId 永远不能授权。
- **subagent 的审批如何路由?** 没有应答者拥有的 agent 穿过整个 waterfall 委派并失败关闭——进程内 subagent 被刻意设计为不可应答。`'never'` 父级会把该覆盖项预置到每个进程内子 agent 的日志中([决策](2026-07-25-subagent-policy-inheritance.md)),因此子 agent 一开始就会得知,而不是向空的 waterfall 发出 ask。`subagent-acp` 的子侧自动应答是独立的;将子 agent 的 ask 路由到父控制器已延后(§ 延后)。
- **subagent 的审批如何路由?** 不路由:委派会把每个进程内子 agent 钉定为 `'never'`([审批钉定决策](2026-08-10-subagent-approval-pinned-never.md)),因此子 agent 的每次 ask 都在任何应答者之前解析为 `rejected`,子 agent 则通过其运行时上下文一开始就会得知。`subagent-acp` 的子侧自动应答是独立的;将子 agent 的 ask 路由到父控制器已延后(§ 延后)。
- **`policy: 'never'` 在运行时实际改变了什么?** 服务在派发任何应答者之前,将该会话的每次 ask 解析为 `rejected`(在服务内部,因此没有注册顺序能绕过它);下一份原子化的运行时上下文快照会声明该策略;每次成功的自动拒绝都会记录审计对。
- **热重载或应答者在会话中途卸载时会发生什么?** 应答者随其拥有的 fiber 一起 dispose,因此下一次 ask 降级为 `unavailable` 而非挂在死通道上;重新挂载会重新注册应答者,无需追赶状态。
- **客户端从哪里获得审批上下文?** 请求携带精确的 `callId` 和发起方的人类可读 `reason`;通道适配器可自行关联更丰富的工具调用状态,而无需在审批 seam 中重复携带参数。

View File

@@ -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 .agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md
2026-07-08-self-referential-cordis-toolset.md: 5fc2fb07fcd0b00bf72c818d3806b298312cdc31
2026-07-08-self-referential-cordis-toolset.zh.md: 8f34c97d94cad9b79a0e823406c07cdcfb38793f
2026-07-08-self-referential-cordis-toolset.md: 0d78e0adff487edae00c2422acc1ef8941e7636a
2026-07-08-self-referential-cordis-toolset.zh.md: 665387a9fb24bf0e6fc04fdfb1ced88b932742ad

View File

@@ -40,7 +40,7 @@ The boundary normalizes unambiguous JSON-Schema forms into `ParameterSchemaSpec`
Every temporary Plugin is a child of one internal `cordis-dynamic` group beneath the tool plugin, so ordinary fiber disposal handles toolset reload and unload. `cordis_mount` awaits settlement; startup failure disposes the fiber before returning an error. A settled pending Plugin remains visible with its missing injections. `cordis_unmount` awaits the Plugin fiber's disposal.
Temporary Plugins exist only in process memory. They create no Plugin file, install no package, change no `cordis.yml` or personal/project configuration, do not survive restart, and have no automatic save, promote, or install path. Keeping an experiment means asking the Agent to implement a normal local, project, or repository Plugin through the regular development workflow.
Temporary Plugins exist only in process memory. They create no Plugin file, install no package, change no `cordis.yml` or personal/project configuration, do not survive restart, and have no automatic save, promote, or install path. Keeping an experiment means asking the Agent to implement a normal project Plugin or installable profile bundle through the regular development workflow.
### Cross-mount composition via provide/inject

View File

@@ -40,7 +40,7 @@ vm 隔离了意外的全局污染,上下文门面隐藏了框架内部细节
每个临时 Plugin 都是工具插件下方内部 `cordis-dynamic` 分组的子节点,因此普通的 fiber 释放即可处理工具集重载和卸载。`cordis_mount` 会等待 settlement;启动失败时在返回错误前释放 fiber。已 settle 但处于 pending 状态的 Plugin 仍然可见,并列出其缺失的注入。`cordis_unmount` 等待 Plugin fiber 的释放完成。
临时 Plugin 只存在于进程内存中。它不会创建 Plugin 文件、安装 package、修改 `cordis.yml` 或个人/项目配置、跨重启存续,也不存在自动保存、转正式或安装路径。若要保留实验结果,应让 Agent 通过常规开发流程实现普通的本地、项目或仓库 Plugin。
临时 Plugin 只存在于进程内存中。它不会创建 Plugin 文件、安装 package、修改 `cordis.yml` 或个人/项目配置、跨重启存续,也不存在自动保存、转正式或安装路径。若要保留实验结果,应让 Agent 通过常规开发流程实现普通的项目 Plugin 或可安装的 profile 组合包。
### 通过 provide/inject 实现跨挂载组合

View File

@@ -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 .agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.md
2026-07-20-dsh-cli-personal-config.md: f5207c5ffbd963b9b7c4a7166fa9f17a460707a9
2026-07-20-dsh-cli-personal-config.zh.md: 24478a4b4fd5878032bf80f5b30ab9e2008da785
2026-07-20-dsh-cli-personal-config.md: 02883c89f27e51d6091d4d65167ebdd6a96f6f51
2026-07-20-dsh-cli-personal-config.zh.md: 6e56e892cf682ea514b036750e44dbb06e944a80

View File

@@ -19,7 +19,7 @@ Two coupled pieces, aligned with the `apps/` assembly tier proposed by the `dsh
**Personal config (`dsh-app-boot`).** The personal overlay lives in the Harness home — `$DSH_HOME`, else `~/.dsh` — resolved by the shared [`resolveDshHome`](../architecture/2026-07-24-single-harness-home-resolver.md) (`@deepseek-ai/dsh-paths`), the same single root skills and AGENTS.md resolve against. The dsh TUI, Web, and headless surfaces consume its two optional files; the demo bins boot their committed trees verbatim:
- `.env` — loaded after the invoking directory's `.env`; `process.loadEnvFile` never overrides, so precedence is ambient > project `.env` > personal `.env`.
- `config.yaml` — a top-level YAML array of `@cordisjs/plugin-include` `PatchOptions`, parsed with the include's own `!!js` dialect (`loadPersonalPatches`) and passed to `boot()`, which forwards it as the root include's `patches`. Patch semantics match the shipped surface overlays: an id-targeted patch replaces the named entry's whole `config`, `insert` appends entries, and an unmatched id is a silent no-op. The [repository Plugin integration](2026-07-30-config-only-repository-plugins.md) uses one shipped row to make an exact GitHub source list a config-only choice.
- `config.yaml` — a top-level YAML array of `@cordisjs/plugin-include` `PatchOptions`, parsed with the include's own `!!js` dialect (`loadPersonalPatches`) and passed to `boot()`, which forwards it as the root include's `patches`. Patch semantics match the shipped surface overlays: an id-targeted patch replaces the named entry's whole `config`, `insert` appends entries, and an unmatched id is a silent no-op. External packages are installed as [profile bundles](../simplification/2026-08-09-remove-repository-plugin.md); this personal layer configures the Loader rows those bundles contribute.
- A missing file means no overlay; a present-but-unreadable, unparsable, or non-array file throws at boot (misconfiguration fails loud, never a silent skip).
The PTY smoke's launcher isolates `$DSH_HOME` to a per-test directory, exactly as it already isolates `DSH_AGENTS_HOME`, so a developer's real personal overlay cannot leak into fixtures; only the dsh CLI reads personal config, so no other test launcher needed changes.
@@ -40,7 +40,7 @@ The TUI and Web register the exact personal path through Cordis HMR after boot.
## Consequences
- `dsh` from any directory (and `pnpm run demo:tui`) can apply personal providers, models, repository Plugins, and other Loader entries with no checkout edit; verified end-to-end against a personal Anthropic proxy with Opus 4.8, including a bash tool round trip.
- `dsh` from any directory (and `pnpm run demo:tui`) can apply personal providers, models, installed bundle entries, and other Loader entries with no checkout edit; verified end-to-end against a personal Anthropic proxy with Opus 4.8, including a bash tool round trip.
- Because an id-targeted patch replaces the whole `config`, a personal override restates the base fields it keeps and can drift when the base entry changes shape; the loader's entry-not-found/name-mismatch warnings and [`dsh --dump-config`](../../../../apps/cli/README.md#profiles) (which prints the composed tree those patches produce) are the diagnostics.
- Personal patches resolve ids against the booted file's own tree, so nested-include overlays (Code Mode) are not personalized; live-run parity for those leaves is deferred.
- `dsh-app-boot` depends on `js-yaml` and imports the include's `!!js` YAML dialect (`entryListSchema`) directly, and, like `apps/cli`, depends on `@deepseek-ai/dsh-paths` for `resolveDshHome`.

View File

@@ -19,7 +19,7 @@ Status: implemented
**个人配置(`dsh-app-boot`)。** 个人 overlay 存放在 Harness home——`$DSH_HOME`,否则 `~/.dsh`——由共享的 [`resolveDshHome`](../architecture/2026-07-24-single-harness-home-resolver.md)(`@deepseek-ai/dsh-paths`)解析,与 skills、AGENTS.md 解析所依据的单一根目录相同。dsh 的 TUI、Web 和无头界面使用其中两个可选文件;各示例 bin 仍然逐字节按已提交的配置树启动:
- `.env`——在调用目录的 `.env` 之后加载;`process.loadEnvFile` 从不覆盖已有值,因此优先级为环境变量 > 项目 `.env` > 个人 `.env`。
- `config.yaml`——顶层 YAML 数组,元素为 `@cordisjs/plugin-include` 的 `PatchOptions`,用 include 自己的 `!!js` 方言解析(`loadPersonalPatches`)并传给 `boot()`,由它作为根 include 的 `patches` 转发。补丁语义与交付的 surface overlay 一致:按 id 定位的补丁替换该配置项的整个 `config`,`insert` 追加配置项,未匹配的 id 静默不执行任何操作。[仓库插件集成](2026-07-30-config-only-repository-plugins.md)通过一个已交付配置项,使精确 GitHub 源列表成为纯配置选择。
- `config.yaml`——顶层 YAML 数组,元素为 `@cordisjs/plugin-include` 的 `PatchOptions`,用 include 自己的 `!!js` 方言解析(`loadPersonalPatches`)并传给 `boot()`,由它作为根 include 的 `patches` 转发。补丁语义与交付的 surface overlay 一致:按 id 定位的补丁替换该配置项的整个 `config`,`insert` 追加配置项,未匹配的 id 静默不执行任何操作。外部包作为 [profile 组合包](../simplification/2026-08-09-remove-repository-plugin.md)安装;这个个人层负责配置这些组合包提供的 Loader 配置项。
- 文件缺失即无 overlay;文件存在但不可读、不可解析或非数组则在启动时抛出(配置错误响亮失败,绝不静默跳过)。
PTY 冒烟测试的启动器把 `$DSH_HOME` 隔离到每个测试自己的目录,与它已有的 `DSH_AGENTS_HOME` 隔离方式完全一致,开发者真实的个人 overlay 不可能泄漏进 fixture;只有 dsh CLI 读取个人配置,因此其他测试启动器无需改动。
@@ -40,7 +40,7 @@ TUI 和 Web 启动后通过 Cordis HMR(热模块替换)注册确切的个人
## Consequences
- 在任意目录运行 `dsh`(以及 `pnpm run demo:tui`),无需修改 checkout,即可应用个人提供方、模型、仓库插件和其他 Loader 配置项;已针对个人 Anthropic 代理与 Opus 4.8 端到端验证,包括一次 bash 工具往返。
- 在任意目录运行 `dsh`(以及 `pnpm run demo:tui`),无需修改 checkout,即可应用个人提供方、模型、已安装组合包的配置项和其他 Loader 配置项;已针对个人 Anthropic 代理与 Opus 4.8 端到端验证,包括一次 bash 工具往返。
- 由于按 id 定位的补丁替换整个 `config`,个人覆盖必须复述它保留的基础字段,并可能随基础配置项形态变化而漂移;诊断手段是 loader 的「配置项未找到/名称不匹配」警告和 [`dsh --dump-config`](../../../../apps/cli/README.md#profiles)(打印这些补丁合成出的配置树)。
- 个人补丁只在被启动文件自身的树里解析 id,因此嵌套 include 的 overlay(Code Mode)不会被个性化;这些叶子的实际运行等价性暂缓。
- `dsh-app-boot` 依赖 `js-yaml`,并直接导入 include 的 `!!js` YAML 方言(`entryListSchema`);与 `apps/cli` 一样依赖 `@deepseek-ai/dsh-paths` 以获取 `resolveDshHome`。

View File

@@ -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 .agents/notes/implemented/feature/2026-07-24-web-session-model-selector.md
2026-07-24-web-session-model-selector.md: e6a96ac62f69a3bd312f61cc920caa259d2dc5b0
2026-07-24-web-session-model-selector.zh.md: 5a0245359d69ac6e59a20dc3276b9411c4b25e23
2026-07-24-web-session-model-selector.md: 3f7dde359842a82d03218dc71ca0e3617ae6dbdf
2026-07-24-web-session-model-selector.zh.md: 474476966da8d4628578dc3e13f74f643bd6eaed

View File

@@ -18,7 +18,7 @@ The browser `ModelService` owns one `ModelDirectory` per live session. Its snaps
`@deepseek-ai/dsh-client-ui-conversation` declares the session-scoped single slot `conversation.input.model` as a child of its composer-bar entry. InputBar renders the seat in its trailing controls immediately before the pending indicator and primary button; the seat receives the bar's `locked` owner prop and session scope. `@deepseek-ai/dsh-client-ui-model` occupies that seat and also contributes `/model` over the same directory. Its compact trigger displays the exact catalog model name and effective reasoning label. When the current selection is absent from the groups, the trigger displays `Select model`, the model list marks no row active, and the Effort row stays absent; choosing a listed model assigns the complete selection through the shared selection path. The upward menu otherwise first offers Model and Effort; Model drills into provider groups, while Effort drills into the adapter-ordered levels. The provider-default row appears only when the adapter does not configure a model default.
The production browser roster is assembled from `apps/cli/config/base.cordis.yml` plus `apps/cli/config/web.cordis.yml`; the model feature is one `dshClient` row rather than a package hardcoded in Web boot code. Its package manifest orders it after the runtime and command feature, while Cordis service injection waits for the conversation slot before registering the composer occupant.
The production browser roster is assembled from `apps/cli/config/base.cordis.yml` plus `apps/cli/config/web.cordis.yml`; the model feature is one `dsh.client` row rather than a package hardcoded in Web boot code. Its package manifest orders it after the runtime and command feature, while Cordis service injection waits for the conversation slot before registering the composer occupant.
## Alternatives considered

View File

@@ -18,7 +18,7 @@ Web Host 为每个新建或恢复的 Agent 安装 `ModelSelection`。如果会
`@deepseek-ai/dsh-client-ui-conversation` 将会话作用域的单实例 slot `conversation.input.model` 声明为其输入栏 entry 的子 slot。InputBar 在尾部控件区将该 seat 渲染于 pending 指示器与主按钮之前;该 seat 接收输入栏的 `locked` owner prop 与会话作用域。`@deepseek-ai/dsh-client-ui-model` 占用该 seat,并在同一目录上提供 `/model`。其紧凑型触发器显示目录中精确模型的名称与生效的推理强度标签。当前选择不在分组中时,触发器显示 `Select model`,模型列表不标记任何活动行,Effort 行也保持隐藏;选择一个已列出的模型,会通过共享的选择路径赋值完整选择。除此情形外,向上展开的菜单会首先提供 Model 与 Effort;Model 可深入提供方分组,Effort 可深入适配器排序的级别。仅当适配器没有配置模型默认值时,才显示提供方默认值行。
生产环境的浏览器名册由 `apps/cli/config/base.cordis.yml` 与 `apps/cli/config/web.cordis.yml` 共同组装;模型功能对应其中一行 `dshClient` 配置项,而不是 Web boot 代码中硬编码的包。其包 manifest(元数据清单)将加载顺序置于运行时与命令功能之后;Cordis 服务注入则等待 conversation slot 可用,再注册 composer 占用方。
生产环境的浏览器名册由 `apps/cli/config/base.cordis.yml` 与 `apps/cli/config/web.cordis.yml` 共同组装;模型功能对应其中一行 `dsh.client` 配置项,而不是 Web boot 代码中硬编码的包。其包 manifest(元数据清单)将加载顺序置于运行时与命令功能之后;Cordis 服务注入则等待 conversation slot 可用,再注册 composer 占用方。
## 考虑过的替代方案

View File

@@ -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 .agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md
2026-07-25-subagent-policy-inheritance.md: aeff83795eedead9c75de6bbb74c1da1945092ca
2026-07-25-subagent-policy-inheritance.zh.md: c26e6bf8b79c86855022c384673957fe04ff761d
2026-07-25-subagent-policy-inheritance.md: 34751a4e29e48c84d37425857b8b1b56c8d866eb
2026-07-25-subagent-policy-inheritance.zh.md: 5fa8edf04ed63da9b2e1b9a062ca2f649c8c96fb

View File

@@ -1,4 +1,4 @@
# Agent Note: In-process subagent policy inheritance — the child starts under the parent's sandbox and approval overrides
# Agent Note: In-process subagent policy inheritance — the child starts under the parent's sandbox override
Status: implemented
@@ -6,11 +6,11 @@ English | [中文](2026-07-25-subagent-policy-inheritance.zh.md)
## Problem
Sandbox and approval overrides are per-session log folds. An in-process subagent gets a new session, so a spawn child once fell back to deployment defaults and a fork child saw only switches inside its completed-turn prefix. Delegation could therefore widen a parent that had switched to `read-only`, or turn a parent's unattended `'never'` approval stance back into prompting behavior.
Sandbox and approval overrides are per-session log folds. An in-process subagent gets a new session, so a spawn child once fell back to deployment defaults and a fork child saw only switches inside its completed-turn prefix. Delegation could therefore widen a parent that had switched to `read-only`.
## Decision
The shared in-process driver snapshots `sandboxPolicy.overrideOf(parent.session)` and `approval.overrideOf(parent.session)` before its first await. A later parent switch belongs to the parent's future; cancel-and-redelegate takes a new snapshot. Both services are optional, and only explicit session overrides are copied, never deployment defaults or one-shot grants.
The delegation boundary snapshots `sandboxPolicy.overrideOf(parent.session)` before its first await, through the shared child-agent helpers (`captureDelegatedPolicyOverrides`/`appendDelegatedPolicyOverrides` in `dsh-subagent`), which the one-shot driver and the [continuable start](2026-08-10-continuable-subagent-policy-inheritance.md) both call. A later parent switch belongs to the parent's future; cancel-and-redelegate takes a new snapshot. The sandbox-policy service is optional, and only the explicit session override is copied, never deployment defaults or one-shot grants. The approval policy is not inherited: the same capture pins every child to `'never'` — the [approvals-pinned decision](2026-08-10-subagent-approval-pinned-never.md) supersedes this note's original approval-override inheritance.
Each captured value becomes a source-tagged `sandbox/mode` or `approval/policy` event appended during the child factory's unpublished setup. The session constructor has already fixed `Session.firstLiveSeq` at the fork-prefix length, so the inherited facts follow fork history, reach telemetry when the child is announced, and leave `SessionHeader.seedLength` at the prefix length. Existing last-event-wins folds therefore make the delegation snapshot beat stale fork history and let a later child switch beat the snapshot. A grandchild folds its parent's logged state, so the rule composes without another inheritance mechanism.
@@ -18,7 +18,7 @@ Ordinary session appends validate the inherited events before publication, and p
### What a blocked child experiences
A confined child gets the ordinary denial marker. No answerer currently owns an in-process child, so an escalation request fails closed and the child reports upward; a controller-owned parent may widen its own session and delegate again. An inherited `'never'` policy tells the child not to request escalation in its first system prompt.
A confined child gets the ordinary denial marker, and an escalation request is rejected deterministically by the child's pinned `'never'` policy; the `subagent:delegation` runtime-context statement tells the child to report the limitation instead of retrying, and a controller-owned parent may widen its own session and delegate again ([approvals-pinned decision](2026-08-10-subagent-approval-pinned-never.md)).
## Alternatives considered
@@ -27,10 +27,10 @@ A confined child gets the ordinary denial marker. No answerer currently owns an
- **A first-prompt listener** — rejected: it introduces listener ordering and a later timing boundary even though the creation transaction already permits log appends before publication.
- **Copying deployment defaults** — rejected: defaults remain operator-owned and may change; an unswitched parent stamps nothing, so its child follows the current deployment.
- **Live resolution walking `parentSession` at each call** — rejected: it breaks the "two sessions never see each other's state" isolation invariant, requires the parent session to stay loaded for the child's lifetime, and makes a mid-run parent switch retroactively change a running child. Snapshot-at-delegation is the semantic: the child keeps the policy it was handed; cancel-and-respawn picks up a tightening.
- **Forcing `'never'` or routing asks to the root controller** — rejected as inheritance behavior. A forced value forecloses a future child answerer; parent routing needs parent-chain ownership and the spawning `callId`, and remains deferred in [the approval-seam Agent Note](2026-07-06-approval-seam.md).
- **Forcing `'never'`** — originally rejected here as inheritance behavior because a forced value forecloses a future child answerer; that verdict is reversed by the [approvals-pinned decision](2026-08-10-subagent-approval-pinned-never.md), which owns the current rationale. Routing asks to the root controller needs parent-chain ownership and the spawning `callId`, and remains deferred in [the approval-seam Agent Note](2026-07-06-approval-seam.md).
## Consequences
- Spawn, fork, and nested in-process children retain a parent's explicit sandbox and approval overrides. The focused suite proves real filesystem denial, stale-fork precedence, delegation-time capture, the live-event boundary, default omission, and context disposal.
- Spawn, fork, and nested in-process children retain a parent's explicit sandbox override and are pinned to `'never'` approvals. The focused suite proves real filesystem denial, stale-fork precedence, delegation-time capture, the live-event boundary, default omission, and context disposal.
- The keyless headless snapshot is the assembled regression: only the parent is `read-only`, the deployment default is `workspace-write`, and the child's persisted event plus denied disk write both fail if capture is removed.
- Each delegation adds at most two log-only events. `dsh-subagent-inprocess` has optional peer types for the two policy services; compositions without either service behave unchanged. Out-of-process children retain their own deployment policy, and a running child does not follow later parent switches.
- Each delegation adds at most two log-only events. `dsh-subagent` owns the optional peer types for the two policy services — its shared helpers hold the `ctx.get` consumption; compositions without either service behave unchanged. Out-of-process children retain their own deployment policy, and a running child does not follow later parent switches.

View File

@@ -1,4 +1,4 @@
# Agent Note: 进程内 subagent 策略继承——子 agent 在父级的沙箱与审批覆盖项下启动
# Agent Note: 进程内 subagent 策略继承——子 agent 在父级的沙箱覆盖项下启动
Status: implemented
@@ -6,11 +6,11 @@ Status: implemented
## 问题
沙箱与审批覆盖项都是按会话的日志折叠。进程内 subagent 会获得一个新会话,因此 spawn 子 agent(智能体)过去会回退到部署默认值,fork 子 agent 则只能看到其已完成轮次前缀中的切换。因此,委派可能放宽已经切换到 `read-only` 的父级,或让父级无人值守的 `'never'` 审批立场重新变成会发起提示的行为。
沙箱与审批覆盖项都是按会话的日志折叠。进程内 subagent 会获得一个新会话,因此 spawn 子 agent(智能体)过去会回退到部署默认值,fork 子 agent 则只能看到其已完成轮次前缀中的切换。因此,委派可能放宽已经切换到 `read-only` 的父级。
## 决策
共享的进程内驱动器在第一次 await 之前对 `sandboxPolicy.overrideOf(parent.session)` 和 `approval.overrideOf(parent.session)` 获取快照。父级后续的切换属于父级的未来;取消后重新委派会取得新快照。这两个服务均为可选,仅复制显式会话覆盖项,绝不复制部署默认值或一次性授权。
委派边界在第一次 await 之前,经由共享的子 agent 辅助函数(`dsh-subagent` 中的 `captureDelegatedPolicyOverrides`/`appendDelegatedPolicyOverrides`)对 `sandboxPolicy.overrideOf(parent.session)` 获取快照;一次性驱动器与[可继续启动](2026-08-10-continuable-subagent-policy-inheritance.md)都会调用这些辅助函数。父级后续的切换属于父级的未来;取消后重新委派会取得新快照。沙箱策略服务为可选,仅复制显式会话覆盖项,绝不复制部署默认值或一次性授权。审批策略不继承:同一次捕获会把每个子 agent 钉定为 `'never'`——[审批钉定决策](2026-08-10-subagent-approval-pinned-never.md)取代了本 note 原先的审批覆盖项继承。
每个捕获值都会成为子 agent 工厂在未发布设置阶段追加的一条带来源标记的 `sandbox/mode` 或 `approval/policy` 事件。会话构造函数已将 `Session.firstLiveSeq` 固定为 fork 前缀的长度,因此继承事实会排在 fork 历史之后,在子 agent 公布时进入遥测,同时让 `SessionHeader.seedLength` 保持为此前缀的长度。因此,既有的末事件胜出折叠会让委派快照压过陈旧的 fork 历史,并让子 agent 后续的切换压过该快照。孙代 agent 会折叠其父级已记录的状态,因此无需另一套继承机制即可组合此规则。
@@ -18,7 +18,7 @@ Status: implemented
### 被拦住的子 agent 会经历什么
受限子 agent 会得到普通拒绝标记。目前没有应答器认领进程内子 agent,因此升级请求会以拒绝方式失败,由子 agent 向上汇报;由控制器持有的父 agent 可以放宽自己的会话后重新委派。继承的 `'never'` 策略会在第一份系统提示词中告知子 agent 不要请求升级。
受限子 agent 会得到普通拒绝标记,升级请求则被子 agent 钉定的 `'never'` 策略确定性拒绝;`subagent:delegation` 运行时上下文声明告知子 agent 上报限制而不是重试,由控制器持有的父 agent 可以放宽自己的会话后重新委派([审批钉定决策](2026-08-10-subagent-approval-pinned-never.md))。
## 考虑过的替代方案
@@ -27,10 +27,10 @@ Status: implemented
- **首个提示词监听器**:不予采纳。尽管创建事务已经允许在发布前追加日志,它仍会引入监听器顺序与更晚的时序边界。
- **复制部署默认值**:不予采纳。默认值仍由运维人员拥有且可能变化;未切换的父级不会记录任何值,因此其子 agent 跟随当前部署。
- **每次调用时沿 `parentSession` 实时解析**:不予采纳。这会打破「两个会话永远看不到彼此状态」的隔离不变量,要求父会话在子 agent 的整个生命周期内保持加载,还会让父级在子 agent 运行途中做的切换追溯性地改变一个正在运行的子 agent。委派时快照才是本设计的语义:子 agent 保持它被交付时的策略;取消后重新 spawn 即可拿到收紧后的策略。
- **强制使用 `'never'` 或把 ask 路由到根控制器**:不作为继承行为采纳。强制值会排除未来的子 agent 应答器;父级路由需要父链所有权与发起 spawn 的 `callId`,仍按[审批 seam Agent Note](2026-07-06-approval-seam.md) 所述延期。
- **强制使用 `'never'`**:本 note 当初不作为继承行为采纳,理由是强制值会排除未来的子 agent 应答器;该结论已被[审批钉定决策](2026-08-10-subagent-approval-pinned-never.md)推翻,现行理由归其所有。把 ask 路由到根控制器需要父链所有权与发起 spawn 的 `callId`,仍按[审批 seam Agent Note](2026-07-06-approval-seam.md) 所述延期。
## 后果
- spawn、fork 和嵌套的进程内子 agent 会保留父级显式的沙箱与审批覆盖项。聚焦测试套件证明真实文件系统拒绝、陈旧 fork 优先级、委派时捕获、实时事件边界、默认值省略与上下文释放。
- spawn、fork 和嵌套的进程内子 agent 会保留父级显式的沙箱覆盖项,并被钉定为 `'never'` 审批。聚焦测试套件证明真实文件系统拒绝、陈旧 fork 优先级、委派时捕获、实时事件边界、默认值省略与上下文释放。
- 无密钥 headless 快照是组装后应用层面的回归测试:只有父级是 `read-only`,部署默认值是 `workspace-write`;若移除捕获,子 agent 的持久化事件与被拒的磁盘写入这两项检查都会失败。
- 每次委派最多增加两条仅日志事件。`dsh-subagent-inprocess` 为两个策略服务提供可选 peer 类型;未组合任一服务的组合保持原有行为。进程外子 agent 仍采用自身的部署策略,正在运行的子 agent 不跟随父级后续切换。
- 每次委派最多增加两条仅日志事件。两个策略服务的可选 peer 类型由 `dsh-subagent` 拥有——其共享辅助函数持有 `ctx.get` 消费;未组合任一服务的组合保持原有行为。进程外子 agent 仍采用自身的部署策略,正在运行的子 agent 不跟随父级后续切换。

View File

@@ -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 .agents/notes/implemented/feature/2026-07-29-persistent-bash-str-replace-editor.md
2026-07-29-persistent-bash-str-replace-editor.md: c4750e30370bfd253064c39cb1adc0f5b2baa60d
2026-07-29-persistent-bash-str-replace-editor.zh.md: 83159d9792fd9fadaaa342cc289300b35da34e4a
2026-07-29-persistent-bash-str-replace-editor.md: 2c077a08e6027245779a0db364c83d17a9c74fce
2026-07-29-persistent-bash-str-replace-editor.zh.md: f642f2100cbc40ddf688400c5e6124ca9a6ff72d

View File

@@ -18,7 +18,7 @@ Some deployments need a one-call Bash schema whose shell state survives across m
Both plugins are included in the Python runtime closure. The persistent Bash closure also includes the PTY service/local backend and the sandbox services required by that backend. Because `node-pty` executes a native `spawn-helper` on macOS, each packaged macOS runtime executable ships with a `-spawn-helper` sibling; Linux uses `forkpty` directly. A pinned `node-pty` patch checks `DSH_NODE_PTY_SPAWN_HELPER` first, so it remains a true override for a current external consumer that supplies a non-sibling helper. When the override is unset, the patch resolves the packaged executable sibling if present and otherwise preserves upstream lookup in ordinary Node runs. The macOS builders fail before publication when the helper is absent or not executable.
The shipped [`core-web.cordis.yml`](../../../../apps/cli/config/core-web.cordis.yml) overlay composes both plugins over the ordinary Web surface for the Claude SWE-compatible RL contract. It pins native tool mode and makes the complete system prompt `DSH_SYSTEM_PROMPT` when set or `You are a helpful software engineer assistant.` otherwise, with no harness identity, source-checkout section, Web orientation, Workspace instructions, or tool-mode guidance. It disables every other model-facing consumer, so the model receives exactly the persistent `bash` and `str_replace_editor` schemas, while the Web host, browser, Workspace, persistence, sandbox, and permission stack remains in place. The local PTY backend resolves the effective session sandbox mode when it creates the shell. While that owner has an open shell or a spawn in progress, a different permission mode is rejected before its session event commits; the editor continues through the Web filesystem sandbox.
The shipped [`minimal` agent preset](../../../../apps/cli/config/agent-presets/minimal/agent.cordis.yml) composes both plugins for the Claude SWE-compatible RL contract. Its entry-local PTY realm carries the registry, local backend, and persistent Bash tool; the editor registers beside that realm against the host filesystem. The preset fixes the complete system prompt, follows the deployment tool-presentation mode, omits every other model-facing consumer, and leaves browser, Workspace, persistence, sandbox, and permission services on the shared Web host. The local PTY backend resolves the effective session sandbox mode when it creates the shell. While that owner has an open shell or a spawn in progress, a different permission mode is rejected before its session event commits; the editor continues through the Web filesystem sandbox. The [minimal-preset decision](../bug-fix/2026-08-10-minimal-preset-owns-rl-composition.md) owns this composition boundary.
## Alternatives considered
@@ -32,4 +32,4 @@ The shipped [`core-web.cordis.yml`](../../../../apps/cli/config/core-web.cordis.
## Consequences
Profiles can reproduce an external agent by configuring persona and descriptions while the underlying packages remain general. Persistent Bash requires an owning Agent and real PTY backend. Shell exit, timeout, or cancellation loses state. The editor delegates security and mutation policy to the mounted filesystem stack. The Core Web profile retains Web permissions but must close its persistent shell before changing modes. Runtime-wheel consumers still need no Node installation; Linux wheels contain one executable, while macOS wheels also contain its private native helper.
Profiles can reproduce an external agent by configuring persona and descriptions while the underlying packages remain general. Persistent Bash requires an owning Agent and real PTY backend. Shell exit, timeout, or cancellation loses state. The editor delegates security and mutation policy to the mounted filesystem stack. A minimal Web agent retains Web permissions but must close its persistent shell before changing modes. Runtime-wheel consumers still need no Node installation; Linux wheels contain one executable, while macOS wheels also contain its private native helper.

View File

@@ -18,7 +18,7 @@ Status: implemented
两个插件都进入 Python runtime 闭包。持久 Bash 的闭包还包含 PTY 服务/本地后端,以及该后端要求的沙箱服务。由于 `node-pty` 在 macOS 上会执行原生 `spawn-helper`,每个打包后的 macOS 运行时可执行文件都会携带一个 `-spawn-helper` 伴随文件;Linux 直接使用 `forkpty`。固定版本的 `node-pty` 补丁会先检查 `DSH_NODE_PTY_SPAWN_HELPER`,因此对当前提供非伴随 helper 的外部消费方而言,该变量仍是真正的覆盖项。未设置该覆盖时,补丁会在打包可执行文件的伴随文件存在时解析它,否则在普通 Node 运行中保留上游查找方式。若 helper 缺失或不可执行,macOS 构建器会在发布前失败。
已交付的 [`core-web.cordis.yml`](../../../../apps/cli/config/core-web.cordis.yml) overlay 会在常规 Web 界面之上组合这两个插件,以满足与 Claude SWE 兼容的 RL 约定。它固定使用原生工具模式;完整的系统提示词在设置 `DSH_SYSTEM_PROMPT` 时采用其值,否则采用 `You are a helpful software engineer assistant.`,且不包含 harness 身份、源码 checkout 提示词段、Web 界面定位、Workspace 指令或工具模式指引。它会禁用其他所有面向模型的消费方,使模型恰好只收到持久 `bash` 和 `str_replace_editor` 两个 schema,同时保留 Web 宿主、浏览器、Workspace、持久化、沙箱与权限栈。本地 PTY 后端会在创建 shell 时解析会话的有效沙箱模式。只要该所有者仍有打开的 shell 或仍在进行中的 spawn,另一种权限模式就会在对应的会话事件提交前遭到拒绝;编辑器则继续经由 Web 文件系统沙箱运行。
随附的 [`minimal` agent preset](../../../../apps/cli/config/agent-presets/minimal/agent.cordis.yml) 会组合这两个插件,以满足与 Claude SWE 兼容的 RL 约定。其 entry 本地 PTY realm 持有注册表、本地后端和持久 Bash 工具;编辑器在该 realm 旁注册,并使用宿主文件系统。preset 会固定完整系统提示词、跟随部署的工具呈现模式,省略其他所有面向模型的消费方,并将浏览器、Workspace、持久化、沙箱与权限服务留在共享 Web 宿主上。本地 PTY 后端会在创建 shell 时解析会话的有效沙箱模式。只要该所有者仍有打开的 shell 或仍在进行中的 spawn,另一种权限模式就会在对应的会话事件提交前遭到拒绝;编辑器则继续经由 Web 文件系统沙箱运行。这一组合边界由 [minimal-preset 决策](../bug-fix/2026-08-10-minimal-preset-owns-rl-composition.md)负责说明。
## 考虑过的替代方案
@@ -32,4 +32,4 @@ Status: implemented
## 后果
Profile 可以通过配置 persona 和描述复现外部 Agent,而底层包保持通用。持久 Bash 需要拥有它的 Agent 与真实 PTY 后端;shell 退出、超时或取消会丢失状态。编辑器把安全与变更策略委托给挂载的文件系统栈。Core Web profile 保留 Web 权限,但必须先关闭持久 shell 才能更改权限模式。运行时 wheel 包的消费方仍无需安装 Node;Linux wheel 包包含一个可执行文件,macOS wheel 包还包含其私有原生 helper。
Profile 可以通过配置 persona 和描述复现外部 Agent,而底层包保持通用。持久 Bash 需要拥有它的 Agent 与真实 PTY 后端;shell 退出、超时或取消会丢失状态。编辑器把安全与变更策略委托给挂载的文件系统栈。minimal Web agent 保留 Web 权限,但必须先关闭持久 shell 才能更改权限模式。运行时 wheel 包的消费方仍无需安装 Node;Linux wheel 包包含一个可执行文件,macOS wheel 包还包含其私有原生 helper。

View File

@@ -1,50 +0,0 @@
# Agent Note: Config-only repository Plugins for standalone dsh
Status: implemented
English | [中文](2026-07-30-config-only-repository-plugins.zh.md)
## Problem
A standalone `dsh` user has no developer-owned SDK project whose `package.json`, lockfile, and `cordis.yml` can carry an external Plugin dependency. Requiring an install command or another state file would make “use this repository” a multi-step workflow, while trusted repository code still needs an exact-source, transactional lifecycle owned by the [repository package format](../architecture/2026-08-08-trusted-repository-package-code.md). Long-running TUI and Web processes also need a failed edit to preserve their usable Plugin generation and tell observers why the candidate was rejected.
## Decision
The shipped TUI and Web/headless `cordis.yml` trees contain an empty `repository-plugins` entry. A user changes only `$DSH_HOME/config.yaml`, replacing that entry's config with a `repositories` list. Each item uses `github:owner/repository#<ref>` plus an optional `&path:/.../.dsh-plugin`; omission selects `/.dsh-plugin`. An explicit ref is mandatory, paths are absolute within the repository and end in `.dsh-plugin`, and duplicate normalized specifiers reject before installation. There is no marketplace, discovery index, HTTPS URL vocabulary, or implicit latest generation.
`@deepseek-ai/dsh-repository-plugin` validates and normalizes each source, then resolves it through the generic vendored [`RepositoryCache`](../architecture/2026-07-30-package-manager-native-repository-cache.md). The default cache is `$DSH_HOME/cache/repository-plugins`; `cacheDir` is the explicit deployment override. Bundled pnpm selects the configured repository subpackage, installs its dependencies, runs its package-authored `prepack`, and atomically publishes the exact specifier. The selected package's direct development dependency on `@deepseek-ai/dsh-repository-plugin` supplies `dsh-plugin-prepare` through package-local `node_modules/.bin`; the lifecycle invokes it after any package-owned build. The DSH host imports the generated `dsh-plugin.mjs` wrapper and mounts it as a child fiber; that wrapper composes static skill and MCP owners plus an explicit trusted Cordis entry when declared.
## Live update and failure
`dsh-app-boot` mounts the root Include through one helper that retains its exact Loader `Entry`. The TUI and Web register `$DSH_HOME/config.yaml` through Cordis HMR; headless reads the same file at startup without retaining a watcher. A watcher update rebuilds the Include patch list as immutable app-owned patches followed by the newly parsed personal patches, so Web-generated port, session-root, trust, and frontend values survive every personal edit unless a later personal patch deliberately replaces that row.
Cordis serializes and coalesces exact-path changes. Include and Loader reconcile a candidate transactionally: success commits the new source list, while fetch, preparation, wrapper import, format, or child-Plugin failure rejects the candidate and retains or restores the last good tree. HMR normalizes the caught value to `Error`, logs it, and broadcasts the parallel `hmr/config-update-failed(filename, error)` event; observer failures cannot break refresh processing. Repository MCP servers use strict startup, so an initial connection, discovery, or tool-registration failure rejects the candidate and becomes a config-update failure; non-strict standalone MCP clients retain their contained successful-Plugin/no-tools behavior.
An identical specifier permanently reuses its cache generation. HMR watches configuration, not cached repository code; the user changes the ref, path, or source list to select another generation.
## Trust boundary
Configuring a repository authorizes package-manager lifecycle code, dependencies, the explicit `dsh.entry`, and spawned MCP servers from that repository to run with the user's filesystem authority. The pnpm child removes ambient environment variables whose names contain `KEY`, `PASSWORD`, `SECRET`, or `TOKEN`, but this is credential-exposure reduction rather than a sandbox. The prepared wrapper validates composition boundaries and lifecycle state; it does not make repository code safe to run when the source is untrusted.
## Alternatives considered
**Require an SDK project dependency.** Rejected for the standalone app path because there is no project manifest to edit. Developer-owned SDK projects keep their native package-manager workflow as a separate capability.
**Add a `dsh plugin install` command and installation database.** Rejected because the personal Loader overlay already owns machine-local composition. A second mutation interface and durable registry would duplicate config identity and rollback.
**Resolve repositories directly in the DSH package.** Rejected because Git transport, GitHub subpackage selection, lifecycle execution, and content storage belong to pnpm and the generic Loader cache, not a DSH-specific adapter.
**Watch cache contents or refresh the same ref automatically.** Rejected because one config value must identify one immutable prepared generation. Background remote resolution would change executable code without a config diff and make rollback depend on mutable remote state.
**Broadcast an `unknown` failure payload.** Rejected at the HMR boundary. JavaScript may throw any value internally, but the public event always receives a normalized `Error`, giving observers one stable contract while retaining the original value as its cause when needed.
## Consequences
- A repository that adds `.dsh-plugin/package.json` can reach standalone users through one personal-config edit without changing its existing skills or `.mcp.json` layout.
- Long-running apps can add, replace, or remove configured generations without restart; rejected candidates retain the last good runtime and produce one generic Cordis event.
- First use may require Git/network access and preparation time. Later starts reuse the exact prepared cache; old generations consume disk until a separate cache-management policy exists.
- Skills and common MCP definitions retain portable static adapters, while an explicit `dsh.entry` can contribute DSH-native Cordis behavior. Format-specific compatibility shims, OAuth-bearing MCP definitions, and marketplaces remain intentionally absent.
## Testing
Repository-package tests pin source normalization, default and nested `.dsh-plugin` paths, cache-root resolution, duplicate rejection, prepared-wrapper loading, and disposal. App-boot tests drive exact-path add, two failure classes, recovery, removal, failure events, and generated-patch preservation through the real HMR/Include/Loader path. A keyless PTY smoke boots the shipped `dsh` composition from personal config alone and invokes a skill from a seeded immutable cache generation.

View File

@@ -1,50 +0,0 @@
# Agent Note: 仅凭配置为独立 dsh 接入仓库插件
Status: implemented
[English](2026-07-30-config-only-repository-plugins.md) | 中文
## 问题
独立 `dsh` 用户没有开发者自有的 SDK 项目,无法由其 `package.json`、lockfile 和 `cordis.yml` 承载外部插件依赖。若要求运行安装命令或维护另一份状态文件,「使用这个仓库」就会变成多步骤流程;受信任的 repository 代码仍需要由[repository 包格式](../architecture/2026-08-08-trusted-repository-package-code.md)负责一套锁定精确来源且具事务性的生命周期。长时间运行的 TUI 和 Web 进程还必须在编辑失败时保留仍可使用的插件版本,并向观察者说明候选配置被拒绝的原因。
## 决策
已交付的 TUI 和 Web/无头 `cordis.yml` 配置树包含一个空的 `repository-plugins` 配置项。用户只需修改 `$DSH_HOME/config.yaml`,用 `repositories` 列表替换该配置项的配置。每一项采用 `github:owner/repository#<ref>`,并可追加 `&path:/.../.dsh-plugin`;省略时选择 `/.dsh-plugin`。必须显式指定 ref;路径是仓库内的绝对路径,并以 `.dsh-plugin` 结尾;重复的规范化说明符在安装前即被拒绝。不提供插件市场、发现索引、HTTPS URL 词汇或隐式的最新版本。
`@deepseek-ai/dsh-repository-plugin` 校验并规范化每个源,再通过 vendor 中的通用 [`RepositoryCache`](../architecture/2026-07-30-package-manager-native-repository-cache.md) 解析。默认缓存位于 `$DSH_HOME/cache/repository-plugins`;`cacheDir` 是显式的部署覆盖项。随应用提供的 pnpm 选择已配置的 repository 子包,安装其依赖,运行包所定义的 `prepack`,并原子发布该精确说明符。所选包对 `@deepseek-ai/dsh-repository-plugin` 的直接开发依赖通过包内 `node_modules/.bin` 提供 `dsh-plugin-prepare`;该生命周期会在任何包自有构建完成后调用它。DSH 宿主会导入生成的 `dsh-plugin.mjs` 包装层并将其挂载为子 fiber;该包装层组合静态 skill(技能)与 MCP 所有者,并在声明时组合显式的受信任 Cordis 入口。
## 实时更新与失败
`dsh-app-boot` 通过一个辅助函数挂载根 Include,并保留其确切的 Loader `Entry`。TUI 和 Web 通过 Cordis HMR(热模块替换)注册 `$DSH_HOME/config.yaml`;无头模式在启动时读取同一文件,但不保留监视器。监视器更新会重新构建 Include 补丁列表,先放置不可变的应用自有补丁,再放置新解析的个人补丁。因此,Web 生成的端口、会话根目录、信任和前端值会在每次个人编辑后保留,除非后续个人补丁有意替换相应配置项。
Cordis 会串行处理并合并该确切路径上的变更。Include 与 Loader 以事务方式协调候选配置:成功时提交新源列表;拉取、准备、包装模块导入、格式或子插件失败时拒绝候选配置,并保留或恢复最后一个可用树。HMR 会把捕获的值规范化为 `Error`,记录错误,并广播并行的 `hmr/config-update-failed(filename, error)` 事件;观察者失败不会中断刷新处理。Repository MCP 服务器采用严格启动,因此初始连接、发现或工具注册失败会拒绝候选配置,并构成配置更新失败;非严格的独立 MCP 客户端仍保留其所收束的「插件成功加载但无工具」行为。
相同说明符会永久复用同一个缓存版本。HMR 监视配置,而非已缓存的仓库代码;用户必须改变 ref、路径或源列表,才能选择另一个版本。
## 信任边界
配置仓库即授权该仓库中的包管理器生命周期代码、依赖、显式 `dsh.entry` 和 spawn 的 MCP server 以用户的文件系统权限运行。pnpm 子进程会移除名称中含有 `KEY`、`PASSWORD`、`SECRET` 或 `TOKEN` 的环境变量,但这只会减少凭据暴露,并非沙箱。已准备的包装层会校验组合边界和生命周期状态;当来源不受信任时,它无法让 repository 代码变得可安全运行。
## 考虑过的替代方案
**要求声明 SDK 项目依赖。** 独立应用路径没有可编辑的项目 manifest(元数据清单),因此否决。开发者自有的 SDK 项目仍可使用原生包管理器工作流,这是一项独立能力。
**新增 `dsh plugin install` 命令和安装数据库。** 否决,因为个人 Loader 覆盖层已经负责机器本地组合。第二个变更接口和持久注册表会重复配置身份与回滚机制。
**由 DSH 包直接解析仓库。** 否决,因为 Git 传输、GitHub 子包选择、生命周期执行和内容存储属于 pnpm 与通用 Loader 缓存,而非 DSH 专用适配器。
**监视缓存内容,或自动刷新相同 ref。** 否决,因为一个配置值必须标识一个不可变的已准备版本。后台远端解析会在没有配置差异的情况下改变可执行代码,并使回滚依赖可变的远端状态。
**广播 `unknown` 失败载荷。** 在 HMR 边界否决。JavaScript 内部可以抛出任意值,但公开事件始终接收规范化的 `Error`,从而为观察者提供稳定约定,并在需要时把原始值保留为错误原因。
## 后果
- 添加 `.dsh-plugin/package.json` 的仓库只需一次个人配置编辑即可供独立用户使用,无需改变现有 skill 或 `.mcp.json` 布局。
- 长时间运行的应用无需重启即可新增、替换或移除已配置版本;被拒绝的候选配置会保留最后一个可用运行时,并产生一个通用 Cordis 事件。
- 首次使用可能需要 Git/网络访问和准备时间。后续启动会复用这份精确的已准备缓存;在另行制定缓存管理政策之前,旧版本会持续占用磁盘空间。
- skill 和通用 MCP 定义保留可移植静态适配器,而显式 `dsh.entry` 可以贡献 DSH 原生 Cordis 行为。格式专用的兼容 shim、带 OAuth 的 MCP 定义和插件市场仍有意不提供。
## 测试
仓库包测试固定源规范化、默认和嵌套 `.dsh-plugin` 路径、缓存根解析、重复项拒绝、已准备包装层加载及资源释放。App-boot 测试通过真实 HMR/Include/Loader 路径驱动确切路径的新增、两类失败、恢复、移除、失败事件及生成补丁保留。一个无密钥 PTY 冒烟测试仅通过个人配置启动已交付的 `dsh` 组合,并从预置的不可变缓存版本中调用一个 skill。

View File

@@ -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 .agents/notes/implemented/feature/2026-07-31-even-out-shipped-tool-rosters.md
2026-07-31-even-out-shipped-tool-rosters.md: 0195620055da5e570d2f54792d950a88bab8d652
2026-07-31-even-out-shipped-tool-rosters.zh.md: ab6982e33c4a0a25cbc2fde386456840ce99d9c5
2026-07-31-even-out-shipped-tool-rosters.md: 7647506e5d9c39d64b686ab18923f9681a48cd87
2026-07-31-even-out-shipped-tool-rosters.zh.md: f04a59c9f00455b00f70975ad9b6bd4defd3d847

View File

@@ -12,11 +12,11 @@ The result was a user-visible difference nobody had decided: the same model, ask
## Decision
The rows that are not surface-specific move into [`base.cordis.yml`](../../../../packages/bundle/base/cordis.patch.yml), and three more join them: `tool-session-query`, `tool-str-replace-editor`, and `repeat-tool-guard`. Web search moves there too; its [deployment decision](2026-07-31-web-default-search.md) owns the security boundary while the shared base owns its surface-neutral mount. Both surfaces assemble the same roster: twenty-two tools on every host — the twenty shared rows plus `glob` and `grep`, which are fixed members because `dsh-tool-fs-search` spawns the [packaged ripgrep binary](../architecture/2026-08-01-packaged-ripgrep-search.md). `tool-session-query` joined and then left again — the [session-search-not-shipped-default decision](2026-08-02-session-search-not-shipped-default.md) keeps the model-facing consumer opt-in — while the rest of this roster stands.
The rows that are not surface-specific move into [`base.cordis.yml`](../../../../packages/bundle/base/cordis.patch.yml), and three more join them: `tool-session-query`, `tool-str-replace-editor`, and `repeat-tool-guard`. Web search moves there too; its [deployment decision](2026-07-31-web-default-search.md) owns the security boundary while the shared base owns its surface-neutral mount. Both surfaces assemble the same roster, including fixed `glob` and `grep` members because `dsh-tool-fs-search` spawns the [packaged ripgrep binary](../architecture/2026-08-01-packaged-ripgrep-search.md). Two later decisions narrow that roster: the [session-search decision](2026-08-02-session-search-not-shipped-default.md) keeps `tool-session-query` opt-in, and the [single-editor decision](../simplification/2026-08-10-default-presets-single-editor.md) keeps `tool-str-replace-editor` out of the general-purpose presets while retaining it in `minimal`.
Two rows stay surface-specific. `tmux-context` is TUI-only because a browser surface has no terminal multiplexer to describe. `session-reference` is TUI-only because it drives the shared session-query index from the launcher's process-local path, and the browser sidebar reconciles that index on its own first search.
**This roster decision added only at the time.** No tool row was removed from either surface when it landed, and a catalog comparison found additions and nothing else. One of those additions, `tool-session-query`, was subsequently removed by the [session-search-not-shipped-default decision](2026-08-02-session-search-not-shipped-default.md). The shared executors, sandbox composition, and access default are owned independently by the [workspace-write default decision](2026-07-31-workspace-write-surface-default.md).
**This roster decision added only at the time.** No tool row was removed from either surface when it landed, and a catalog comparison found additions and nothing else. The later session-search and single-editor decisions own their respective default-roster exceptions. The shared executors, sandbox composition, and access default are owned independently by the [workspace-write default decision](2026-07-31-workspace-write-surface-default.md).
### What stays unmounted, and why

View File

@@ -12,11 +12,11 @@ Status: implemented
## 决策
那些并非 surface 专属的行移入 [`base.cordis.yml`](../../../../packages/bundle/base/cordis.patch.yml),另有三行加入:`tool-session-query`、`tool-str-replace-editor` 和 `repeat-tool-guard`。Web 搜索也一并移入;其[部署决策](2026-07-31-web-default-search.md)负责安全边界,共享 base 则负责与 surface 无关的挂载。两个 surface 组装同一份清单:每台宿主上都有二十二个工具——二十个共享行加上 `glob` 和 `grep`,它们成为固定成员,因为 `dsh-tool-fs-search` 直接 spawn [打包的 ripgrep 二进制](../architecture/2026-08-01-packaged-ripgrep-search.md)。`tool-session-query` 加入后又退出了——[session-search-not-shipped-default 决策](2026-08-02-session-search-not-shipped-default.md)让面向模型的消费方保持需显式启用——而这份清单的其余部分保持不变。
那些并非 surface 专属的行移入 [`base.cordis.yml`](../../../../packages/bundle/base/cordis.patch.yml),另有三行加入:`tool-session-query`、`tool-str-replace-editor` 和 `repeat-tool-guard`。Web 搜索也一并移入;其[部署决策](2026-07-31-web-default-search.md)负责安全边界,共享 base 则负责与 surface 无关的挂载。两个 surface 组装同一份清单,其中 `glob` 和 `grep` 是固定成员,因为 `dsh-tool-fs-search` 直接 spawn [打包的 ripgrep 二进制](../architecture/2026-08-01-packaged-ripgrep-search.md)。之后有两项决策收窄这份清单:[session-search 决策](2026-08-02-session-search-not-shipped-default.md)让 `tool-session-query` 保持需显式启用,[单一编辑器决策](../simplification/2026-08-10-default-presets-single-editor.md)让通用 preset 不提供 `tool-str-replace-editor`,但在 `minimal` 中保留它。
有两行仍是 surface 专属。`tmux-context` 只在 TUI,因为浏览器 surface 没有终端复用器可描述。`session-reference` 只在 TUI,因为它以 launcher 的进程本地路径驱动共享的 session-query 索引,而浏览器侧边栏会在自己的首次搜索里重建该索引。
**本次工具清单决策当时只做加法。** 落地时两个 surface 均未移除任何工具行,目录对比只发现了新增,别无其他。这些新增中的一项 `tool-session-query` 随后被[session-search-not-shipped-default 决策](2026-08-02-session-search-not-shipped-default.md)移除。共享执行器、沙箱组合与访问默认值独立归属[workspace-write 默认值决策](2026-07-31-workspace-write-surface-default.md)。
**本次工具清单决策当时只做加法。** 落地时两个 surface 均未移除任何工具行,目录对比只发现了新增,别无其他。后续的 session-search 与单一编辑器决策分别负责对应的默认清单例外。共享执行器、沙箱组合与访问默认值独立归属[workspace-write 默认值决策](2026-07-31-workspace-write-surface-default.md)。
### 什么保持不挂,以及为什么

View File

@@ -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 .agents/notes/implemented/feature/2026-08-02-session-search-not-shipped-default.md
2026-08-02-session-search-not-shipped-default.md: 65bd72fff76210b726e7562fb8e88e5f8802434a
2026-08-02-session-search-not-shipped-default.zh.md: 4eb0851c1e584b84847b6bb5118c8bb2f3156845
2026-08-02-session-search-not-shipped-default.md: c1bfd7f8e354a4480c5635619514fe782ea71d2c
2026-08-02-session-search-not-shipped-default.zh.md: 9b80c549425c26055700480dd57f1a0a7d01e4a8

View File

@@ -10,7 +10,7 @@ The [shipped-roster decision](2026-07-31-even-out-shipped-tool-rosters.md) made
## Decision
The shipped TUI, Web, and headless surfaces no longer mount `@deepseek-ai/dsh-tool-session-query`: the row is removed from the shared `cordis.patch.yml`, the now-dangling `disabled` patch in the opt-in [`core-web.cordis.yml`](../../../../apps/cli/config/core-web.cordis.yml) profile goes with it, and the workspace dependency drops from `apps/cli/package.json`. The consumer stays opt-in exactly as the model-facing-session-query-tools note describes: the ACP example's [`session-query.cordis.yml`](../../../../examples/acp-agent/session-query.cordis.yml) and its snapshot counterpart remain the mounted reference, and a custom composition can mount the package with the timeout and spill policies.
The shipped TUI, Web, and headless surfaces do not mount `@deepseek-ai/dsh-tool-session-query`, and no shipped agent preset carries it. The consumer stays opt-in exactly as the model-facing-session-query-tools note describes: the ACP example's [`session-query.cordis.yml`](../../../../examples/acp-agent/session-query.cordis.yml) and its snapshot counterpart remain the mounted reference, and a custom composition can mount the package with the timeout and spill policies.
The `ctx.sessionQuery` service itself stays mounted. `session-query-sqlite` remains a base row — the TUI's `session-reference` consumes it for `/resume` — and the Web overlay keeps patching it to an in-memory index for the browser content search. Only the model-facing consumer is removed.

View File

@@ -10,7 +10,7 @@ Status: implemented
## 决策
交付的 TUI、Web 与无头 surface 不再挂载 `@deepseek-ai/dsh-tool-session-query`:该行从共享的 `cordis.patch.yml` 移除,opt-in 的 [`core-web.cordis.yml`](../../../../apps/cli/config/core-web.cordis.yml) profile 中那条已悬空的 `disabled` patch 也随之删除,workspace 依赖也从 `apps/cli/package.json` 中移除。该消费方仍保持 opt-in,与面向模型的会话查询工具决策所述完全一致:ACP(Agent Client Protocol)示例的 [`session-query.cordis.yml`](../../../../examples/acp-agent/session-query.cordis.yml) 及其快照对侧文件仍是挂载参考,自定义组合也可以连同超时与 spill 策略一起挂载该包。
交付的 TUI、Web 与无头 surface 均不挂载 `@deepseek-ai/dsh-tool-session-query`,交付的 agent preset 也都不包含它。该消费方仍保持 opt-in,与面向模型的会话查询工具决策所述完全一致:ACP(Agent Client Protocol)示例的 [`session-query.cordis.yml`](../../../../examples/acp-agent/session-query.cordis.yml) 及其快照对侧文件仍是挂载参考,自定义组合也可以连同超时与 spill 策略一起挂载该包。
`ctx.sessionQuery` 服务本身保持挂载。`session-query-sqlite` 仍是 base 的一行,TUI 的 `session-reference` 消费它来实现 `/resume`,Web overlay 也继续把它 patch 成内存索引,供浏览器内容搜索使用。被移除的只有面向模型的消费方。

View File

@@ -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 run verify-translation-pairing --write .agents/notes/implemented/feature/2026-08-10-continuable-subagent-policy-inheritance.md
2026-08-10-continuable-subagent-policy-inheritance.md: c9b75f2840eb2f124f040d138b761ee145fc6f83
2026-08-10-continuable-subagent-policy-inheritance.zh.md: 8bd7f68c578ed827c756a415f017eb8cb61e5721

View File

@@ -0,0 +1,29 @@
# Agent Note: Continuable subagent policy inheritance — the durable child log owns the delegation-time snapshot
Status: implemented
English | [中文](2026-08-10-continuable-subagent-policy-inheritance.zh.md)
## Problem
The one-shot in-process driver has seeded parent sandbox/approval overrides into its children since the [in-process policy-inheritance decision](2026-07-25-subagent-policy-inheritance.md), but the continuable path never did: `SubagentContinuationManager` materialization applied only child composition and the activation setup registry. The default bundle wires both delegation tools as `backgroundMode: continuable`, so in a default deployment every background child silently fell back to deployment defaults — a parent switched to `danger-full-access` produced children stuck at `workspace-write` whose every out-of-workspace operation raised an approval prompt, and a parent's unattended `'never'` approval stance reverted to prompting ([dsh-external/issues#334](https://github.com/dsh-external/issues/issues/334)).
## Decision
The capture/append pair moved from the one-shot driver into the seam's shared child-agent module (`dsh-subagent/src/child-agent.ts`), the declared one home for shared child composition: `captureDelegatedPolicyOverrides(parent)` snapshots `sandboxPolicy.overrideOf(parent.session)` through optional `ctx.get` and pins the child approval policy to `'never'` ([approvals-pinned decision](2026-08-10-subagent-approval-pinned-never.md)), and `appendDelegatedPolicyOverrides(childSession, overrides)` appends the `source: 'delegation'` events. The one-shot driver and the continuation manager both call them, so the two paths cannot drift.
`startContinuable` captures before its first await (`prepareContinuable`), the same "a later parent switch belongs to the parent's future" boundary as one-shot. The snapshot travels in `MaterializeInputs.create`, so only fresh materialization appends the events during unpublished setup, after any fork seed. A cold resume passes no `create` inputs and appends nothing: the persisted child log already carries the delegation events, and replaying the log IS the state. The durable child log — not the current Activation, not the resuming parent — owns the child's effective policy, so a parent switch between residency epochs never retroactively changes a durable child.
## Alternatives considered
- **An activation-setup-registry contribution** (`registerContinuableSetup`) — rejected: a contribution receives only the child context, so it cannot capture the parent's overrides at the delegation boundary; the registry applies on cold resume as well as fresh creation, which would re-append or re-capture; and nothing ties a contribution's capture to the start call's synchronous prefix, so the pre-await capture guarantee would be lost.
- **Re-capturing the parent's overrides at cold resume** — rejected: a resumed child would silently change policy with the parent's later switches, breaking the snapshot-at-delegation semantic and making effective policy depend on resume timing instead of the child's own log. A parent that wants a resumed child under new policy re-delegates.
- **Importing the one-shot driver's inline logic from the continuation manager** — rejected: the Service Definition package cannot depend on its own provider package, and duplicating the capture/append pair in `continuation.ts` invites drift; `child-agent.ts` already holds every other shared composition step.
- **Seeding the events into the descriptor seed turn** — rejected: the capture value is not known when the seed is assembled for every caller, and the one-shot precedent already establishes unpublished-setup appends as the ordering that places inherited facts after fork history with `firstLiveSeq` intact.
## Consequences
- Default-bundle background delegation (`backgroundMode: continuable`) now inherits a parent's explicit sandbox override and pins the child to `'never'` approvals; compositions without either policy service behave unchanged.
- `dsh-subagent` gains optional peer types on `dsh-sandbox-policy` and `dsh-user-approval` (the `ctx.get` pattern the one-shot driver used); `dsh-subagent-inprocess` drops its policy-service peers and type imports entirely and delegates to the shared helpers.
- The continuable suite (`packages/subagent/subagent/tests/continuation-inheritance.spec.ts`) pins fresh-start seeding, pre-await capture, default omission, cold-resume snapshot stability, and fork-seed precedence; the ACP snapshot scenario `subagent-continuable-inheritance` pins the child's delegation event and read-only runtime context through the assembled app and fails when the capture is removed.
- Out-of-process providers (`acp`, `dsh-sdk`, `claude-code`, `codex`) support no continuable children (`prepareContinuable` absent), and their one-shot children keep their own deployment policy (`inheritsParentContext = false`); cross-process policy propagation remains out of scope.

View File

@@ -0,0 +1,29 @@
# Agent Note: 可继续 subagent 策略继承——持久化子日志拥有委派时快照
Status: implemented
[English](2026-08-10-continuable-subagent-policy-inheritance.md) | 中文
## 问题
自[进程内策略继承决策](2026-07-25-subagent-policy-inheritance.md)以来,一次性进程内驱动器一直会把父级的沙箱/审批覆盖项注入其子级,但可继续路径从未这样做:`SubagentContinuationManager` 的物化只应用子级组合与 Activation(激活)设置注册表。默认组合包把两个委派工具都配置为 `backgroundMode: continuable`,因此在默认部署中,每个后台子 agent(智能体)都静默回退到部署默认值:切换到 `danger-full-access` 的父级产出的子 agent 卡在 `workspace-write`,每次工作区外操作都会触发审批提示;父级无人值守的 `'never'` 审批立场也退回为发起提示的行为([dsh-external/issues#334](https://github.com/dsh-external/issues/issues/334))。
## 决策
捕获/追加这对函数从一次性驱动器移入该 seam 的共享子 agent 模块(`dsh-subagent/src/child-agent.ts`),即声明的共享子级组合唯一归属之处:`captureDelegatedPolicyOverrides(parent)` 通过可选的 `ctx.get` 对 `sandboxPolicy.overrideOf(parent.session)` 建立快照,并把子级审批策略钉定为 `'never'`([审批钉定决策](2026-08-10-subagent-approval-pinned-never.md)),`appendDelegatedPolicyOverrides(childSession, overrides)` 则追加 `source: 'delegation'` 事件。一次性驱动器与继续执行管理器都调用它们,因此两条路径不会出现偏差。
`startContinuable` 在其第一次 await(`prepareContinuable`)之前完成捕获,沿用与一次性路径相同的「父级后续切换属于父级的未来」边界。快照放在 `MaterializeInputs.create` 中传递,因此只有全新物化会在未发布的设置阶段、排在任何 fork 种子之后追加这些事件。冷恢复(cold resume)不传入 `create` 输入,也不追加任何内容:持久化的子日志已经携带委派事件,而回放该日志本身就是状态。子 agent 的生效策略由持久化子日志拥有,而不是当前 Activation,也不是发起恢复的父级,因此父级在驻留纪元(residency epoch)之间的切换绝不会追溯性地改变一个持久化子 agent。
## 考虑过的替代方案
- **一项 Activation 设置注册表贡献**(`registerContinuableSetup`):不予采纳。贡献只接收子级上下文,因此无法在委派边界捕获父级的覆盖项;该注册表在冷恢复与全新创建时都会应用,会导致重复追加或重复捕获;而且没有任何机制把贡献的捕获绑定到 start 调用的同步前缀,await 前捕获的保证会因此丢失。
- **在冷恢复时重新捕获父级覆盖项**:不予采纳。恢复的子 agent 会随父级后续切换静默改变策略,这会破坏委派时快照的语义,并让生效策略取决于恢复时机而非子级自身的日志。希望恢复的子 agent 采用新策略的父级应重新委派。
- **让继续执行管理器导入一次性驱动器的内联逻辑**:不予采纳。Service Definition 包不能依赖自己的提供方包,而在 `continuation.ts` 中复制捕获/追加这对函数会招致偏差;`child-agent.ts` 已经承载其余每个共享组合步骤。
- **把这些事件写入描述符种子轮次**:不予采纳。种子为每个调用方组装时,捕获值尚不可知;而且一次性路径的先例已经确立:在未发布的设置阶段追加,才是把继承事实排在 fork 历史之后、同时保持 `firstLiveSeq` 不变的顺序。
## 后果
- 默认组合包的后台委派(`backgroundMode: continuable`)现在会继承父级显式的沙箱覆盖项,并把子级钉定为 `'never'` 审批;未组合任一策略服务的组合保持原有行为。
- `dsh-subagent` 新增针对 `dsh-sandbox-policy` 与 `dsh-user-approval` 的可选 peer 类型(即一次性驱动器所用的 `ctx.get` 模式);`dsh-subagent-inprocess` 完全移除自己的策略服务 peer 与类型导入,委托给共享辅助函数。
- 可继续测试套件(`packages/subagent/subagent/tests/continuation-inheritance.spec.ts`)锁定全新启动的种子写入、await 前捕获、默认值省略、冷恢复快照稳定性与 fork 种子优先级;ACP 快照场景 `subagent-continuable-inheritance` 经组装后的应用锁定子级的委派事件与只读运行时上下文,移除捕获时即失败。
- 进程外提供方(`acp`、`dsh-sdk`、`claude-code`、`codex`)不支持可继续子 agent(没有 `prepareContinuable`),其一次性子 agent 保留自身的部署策略(`inheritsParentContext = false`);跨进程策略传播仍不在范围内。

View File

@@ -1,6 +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 run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-30-config-only-repository-plugins.md
2026-07-30-config-only-repository-plugins.md: 35327a30e03c51311f634e05ade209ab93ae0155
2026-07-30-config-only-repository-plugins.zh.md: 5755045560da761b59f7c65e99d551f599c2b5b3
# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-08-10-subagent-approval-pinned-never.md
2026-08-10-subagent-approval-pinned-never.md: a21c6b966b1ad00ed63e0fe87b0ce982f0daf490
2026-08-10-subagent-approval-pinned-never.zh.md: db44ae134d34904a53691cfe78eaa5a899cf64e0

View File

@@ -0,0 +1,34 @@
# Agent Note: Delegated subagents run with approvals pinned to `'never'`
Status: implemented
English | [中文](2026-08-10-subagent-approval-pinned-never.zh.md)
## Problem
A delegated child that asked for approval had no one to ask. Under an interactive parent (`'ask'`), a background child's escalation became a pending question no product surface showed — subagent sessions are omitted from the Web sidebar, the parent's `list_agents` reports plain `running`/`idle`, and the catalog rows show only activity — so a permission-blocked child was indistinguishable from a working one; headless and unanswered compositions failed the same ask closed as `'unavailable'`. The rejection audit landed only in the child's own log, and no tool parameter or Web control can adjust a running child session's sandbox mode or approval policy (Issue #1723). The mechanism-heavy fix — a durable blocked-state projection, parent notices, catalog badges, and a permission write path through the subagent ownership fence — was disproportionate directly before release.
## Decision
A delegated child acts only within the permission scope fixed at delegation, and approval prompts are removed from its world entirely: `captureDelegatedPolicyOverrides(parent)` (`dsh-subagent/src/child-agent.ts`) still snapshots the parent session's explicit sandbox override, but pins `approvalPolicy: 'never'` whenever the approval capability is composed — it no longer reads the parent's own approval policy. `appendDelegatedPolicyOverrides()` writes the pin as the durable `approval/policy { policy: 'never', source: 'delegation' }` event on the child's log, through the same one-shot and continuable delegation paths as the sandbox snapshot, so cold resume replays it and a fork seed's stale parent policy loses to it.
Enforcement is the existing `ApprovalService` `'never'` semantics at the one operation that decides asks: every child ask — a `sandbox_permissions` escalation from bash or fs, a hook-driven permission question, any future asker — resolves `'rejected'` deterministically before any answerer is consulted, still leaving the `approval/asked`/`approval/decided` audit pair on the child log. The child's whole permission story is therefore its sandbox scope: a `danger-full-access` parent delegates children that need no approvals, a `read-only` parent delegates children with no escape hatch, and a widening decision always belongs to the parent side (widen the parent session, then delegate or follow up again).
Every in-process child is told, not trapped: `applyChildComposition` registers the scoped `subagent:delegation` runtime-context statement (order 120, after the `sandbox:policy` and `approval:policy` sentences) stating that the scope was fixed at start, approval-requiring operations are rejected automatically, and a task needing wider access ends with a reported limitation instead of retries. The statement is a runtime-context contribution rather than a system-prompt section, so the deployment's system prompt stays uniform across parents and children (the snapshot suite pins that uniformity) and the fact rides the same durable snapshot as the policy sentences.
This supersedes the approval half of the [in-process delegation-policy decision](2026-07-25-subagent-policy-inheritance.md) and reverses its "forcing `'never'` forecloses a future child answerer" verdict: approval inheritance shipped, produced the invisible blocked states above, and a future child answerer now requires reversing this note first.
## Alternatives considered
- **Inheriting the parent's approval override** (the prior behavior) — rejected: only a parent already at `'never'` produced deterministic children; an interactive parent seeded children whose asks waited on a prompt no one was watching or failed closed `'unavailable'`, and the outcome depended on which surfaces happened to be attached.
- **Blocked-state visibility and per-child permission adjustment** (the original #1723 acceptance) — deferred, not rejected: a `list_agents` blocked annotation, parent notices over the settlement-delivery seam, catalog badges, and a subagent-routed permission channel remain the richer design, but each needs its own seam work and none is required once children cannot enter a blocked-waiting state.
- **Routing child asks to the parent controller** — still deferred in the [approval-seam Agent Note](2026-07-06-approval-seam.md): it needs parent-chain ownership and the spawning `callId`.
- **Pinning inside `ApprovalService` by session origin** — rejected: it couples the approval package to delegation vocabulary and duplicates a decision the delegation boundary already owns; the delegation-seeded event is enforceable because no current write path can switch a child session's policy (the `/permission` command requires generic Host routing, which the subagent ownership fence denies to child sessions).
## Consequences
- The child's sandbox inheritance is the complete delegation permission model; the `DelegatedPolicyOverrides.approvalPolicy` field narrows to `'never' | undefined` (`undefined` only without a composed approval capability).
- Model-visible: each child's runtime-context snapshot carries the `subagent:delegation` statement plus the standing disabled-approvals sentence; parent requests are unchanged. The executor-boundary test proves a child escalation is rejected without consulting a root answerer that would have granted it, with the audit pair logged.
- Boundaries: in-process one-shot, continuable, and workflow-spawned children are enforced through the shared helpers; `subagent-acp` children keep that provider's explicit machine `permission` policy; `claude-code`, `codex`, and `dsh-sdk` children run in external processes under their own composition.
- Children persisted before the pin fold to the deployment approval default on cold resume; pre-release, no migration is added.
- Snapshot fixtures record the pin: every in-process child log gains the delegation `approval/policy` event, and `subagent-published-run-failure` now persists a one-event child log where the child previously left no durable events.

View File

@@ -0,0 +1,34 @@
# Agent Note: 被委派的 subagent 以钉定为 `'never'` 的审批策略运行
Status: implemented
[English](2026-08-10-subagent-approval-pinned-never.md) | 中文
## 问题
被委派的子 agent 发起审批请求时无人可问。在交互式父级(`'ask'`)之下,后台子 agent 的升级请求会变成一个任何产品界面都不展示的挂起问题——subagent 会话不进入 Web 侧边栏,父级的 `list_agents` 只报告普通的 `running`/`idle`,目录树的行也只显示活动状态——因此被权限拦住的子 agent 与正常干活的子 agent 无法区分;headless 与无应答者的组合则让同一次 ask 以 `'unavailable'` 失败关闭。拒绝的审计记录只落在子 agent 自己的日志里,而且没有任何工具参数或 Web 控件能调整一个正在运行的子会话的沙箱模式或审批策略(Issue #1723)。机制繁重的修复方案——持久化的受阻状态投影、父级通知、目录树徽标,以及穿过 subagent 所有权围栏的权限写入路径——在临近发布时代价不成比例。
## 决策
被委派的子 agent 只在委派时固定的权限范围内行动,审批提示则从它的世界中彻底移除:`captureDelegatedPolicyOverrides(parent)`(`dsh-subagent/src/child-agent.ts`)仍对父会话的显式沙箱覆盖项建立快照,但只要审批能力已组合,就把 `approvalPolicy: 'never'` 钉定下来——不再读取父级自身的审批策略。`appendDelegatedPolicyOverrides()` 把这个钉定作为持久化的 `approval/policy { policy: 'never', source: 'delegation' }` 事件写入子 agent 的日志,与沙箱快照走完全相同的一次性与可继续委派路径,因此冷恢复会重放它,fork 种子中陈旧的父级策略也会输给它。
强制执行沿用既有的 `ApprovalService` `'never'` 语义,落在裁决 ask 的唯一操作上:子 agent 的每次 ask——bash 或 fs 的 `sandbox_permissions` 升级、hook 驱动的权限询问、任何未来的请求方——都在咨询任何应答者之前确定性地解析为 `'rejected'`,同时仍在子日志上留下 `approval/asked`/`approval/decided` 审计对。子 agent 的全部权限故事因此就是它的沙箱范围:`danger-full-access` 父级委派出的子 agent 无需任何审批,`read-only` 父级委派出的子 agent 没有任何逃生通道,而放宽的决定始终属于父级一侧(先放宽父会话,再重新委派或继续 follow-up)。
每个进程内子 agent 都被告知而非被困住:`applyChildComposition` 注册作用域内的 `subagent:delegation` 运行时上下文声明(order 120,位于 `sandbox:policy` 与 `approval:policy` 语句之后),声明权限范围已在启动时固定、需要审批的操作会被自动拒绝、需要更宽访问的任务应以上报限制收尾而不是重试。该声明是运行时上下文贡献而非系统提示词 section,因此部署的系统提示词在父子之间保持统一(快照测试套件钉住了这一统一性),该事实也随策略语句乘坐同一份持久化快照。
本决策取代[进程内委派策略决策](2026-07-25-subagent-policy-inheritance.md)中的审批一半,并推翻其「强制 `'never'` 会排除未来的子 agent 应答器」的结论:审批继承已经落地,产生的正是上述不可见的受阻状态;未来若要引入子 agent 应答器,必须先推翻本 note。
## 考虑过的替代方案
- **继承父级的审批覆盖项**(先前的行为):不予采纳。只有已处于 `'never'` 的父级才产生确定性的子 agent;交互式父级种出的子 agent,其 ask 要么等待一个无人在看的提示,要么以 `'unavailable'` 失败关闭,结果取决于当时恰好接入了哪些界面。
- **受阻状态可见性与逐子级权限调整**(#1723 原有的验收):延后而非否决。`list_agents` 的受阻标注、经由结算投递 seam 的父级通知、目录树徽标,以及 subagent 专用的权限通道仍是更完整的设计,但每一项都需要独立的 seam 工作;一旦子 agent 不可能进入等待审批的受阻状态,这些都不再是必需。
- **把子 agent 的 ask 路由到父控制器**:仍按[审批 seam Agent Note](2026-07-06-approval-seam.md) 延后。它需要父链所有权与发起 spawn 的 `callId`。
- **在 `ApprovalService` 内按会话来源钉定**:不予采纳。这会让审批包耦合委派词汇,并重复一个委派边界已经拥有的决定;委派种入的事件之所以可强制执行,是因为当前不存在任何能切换子会话策略的写入路径(`/permission` 命令要求通用 Host 路由,而 subagent 所有权围栏对子会话拒绝该路由)。
## 后果
- 子 agent 的沙箱继承就是委派权限模型的全部;`DelegatedPolicyOverrides.approvalPolicy` 字段收窄为 `'never' | undefined`(仅在未组合审批能力时为 `undefined`)。
- 模型可见:每个子 agent 的运行时上下文快照携带 `subagent:delegation` 声明以及固定的审批已禁用语句;父级请求不变。executor 边界测试证明:即使根部有一个本会批准的应答者,子 agent 的升级仍被拒绝且不咨询该应答者,审计对照常落日志。
- 边界:进程内一次性、可继续以及 workflow 派生的子 agent 都经由共享辅助函数强制执行;`subagent-acp` 子 agent 保留该提供方显式的机器 `permission` 策略;`claude-code`、`codex` 与 `dsh-sdk` 子 agent 运行在外部进程中,由各自的组合决定。
- 在钉定之前持久化的子 agent 冷恢复时折叠到部署审批默认值;处于预发布阶段,不添加迁移。
- 快照夹具记录了该钉定:每个进程内子日志都新增委派 `approval/policy` 事件,`subagent-published-run-failure` 现在会持久化一份单事件子日志,而此前该子 agent 不留任何持久化事件。