@deepseek-ai/dsh-session 0.1.2-rc.1 → 0.1.5-alpha.1

package/README.i18n.yaml

@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/core/session/README.md
-README.md: f5cf910854203021a619cc786dfd13705927ffc1
-README.zh.md: 385d7fd63a6e4dec9c23c9d38a352942d7dbc7f9
+README.md: a06b22b5e8f48743047c68e883edd496fbc6eb77
+README.zh.md: 755894b4c30f0a3807f472cece32685d102efbeb

package/README.md

@@ -9,7 +9,7 @@ English | [中文](README.zh.md)
 
 ## Summary
 
-`dsh-session` provides the append-only session log that records an agent's whole interaction history — the single source of truth every model-visible fact flows through. The LLM message history is *derived* from the log (`deriveMessages()`), never stored separately, so replay is re-derivation from the same events and compaction can shadow older surface entries without deleting history. The package also provides the in-memory store (`ctx.sessions`), the typed `SessionEvent` vocabulary that plugins extend by declaration merging, and the surface layer that orders message-producing events. Persistence is deliberately a separate concern: backends subscribe to `session/event` and flush on `session/flush`. Choose it as the foundation of any agent session; it runs no model calls itself.
+`dsh-session` records every model-visible fact in an append-only session log and derives model history from that record. Consumers can inspect, replay, fork, and flush sessions while preserving historical events; compaction hides superseded entries from the active conversation without deleting them. Sessions remain in memory unless a persistence backend is added, and durability checkpoints wait for configured backends. Choose this package wherever an agent needs a reconstructable session record; it does not call models.
 
 ## Table of Contents
 
@@ -47,7 +47,11 @@ session.append('user/message', { role: 'user', content: [{ type: 'text', text: '
 session.deriveMessages()         // the derived model history
 ```
 
-Surface events (`user/message`, `assistant/message`, `tool/result`) must declare how they join the ordered surface; raw chunks, boundaries, and other log-only events never produce a message.
+Surface events (`system/message`, `user/message`, `assistant/message`, `tool/result`) require `surfaceOp` in both typed events and append input. A replacement uses exactly `{ op: 'replace', startSeq, endSeq }`, with inclusive `SessionSeq` endpoints in current surface order. An Assistant message embeds its exact compact provider stream and forbids `sourceEventSeqs`. Known log-only events forbid both metadata fields and never produce a message.
+
+Append, seed/restore, and event adoption/snapshot reject any `header.system` and exactly empty optional request-header fields (`tools: []`, `adapterDefaults: {}`) instead of normalizing input. Tool-result `data.error` is allowed only when `message.content[0].isError === true`; failure identity remains optional. Rejected appends do not change the log, derived state, or event feed. Adoption validates event-local metadata but not referenced history or replacement membership.
+
+`system/message` holds the rendered system prompt: the first one is surface node 0, the prepared call capability governs admission, with a non-empty rendering consolidated at the first system node on an incapable route or appended after cached history inside a continuing `in-history` series; empty system nodes project to no message, so clearing the prompt requires logged empty replacements of all active system nodes, not just the latest; the surface fold rejects a replacement covering node 0 while it is a `system/message` unless the replacing event is itself a `system/message` over exactly that node, while later system nodes carry no protection and a compaction range may shadow them ([decision](../../../.agents/notes/implemented/architecture/2026-09-02-system-prompt-as-surface-node.md)).
 
 ### Read the log
 
@@ -77,7 +81,7 @@ This section explains how the package realizes the behavior above; the observabl
 
 ### Design concept
 
-The package is built on event sourcing: a `Session` is an append-only log of typed `SessionEvent`s, and everything else — model history, transcripts, telemetry, titles, persistence — derives from that stream. The surface is a derived projection: an incremental manager validates append candidates, advances the ordered view from committed events, and tracks a `replaceGeneration` that bumps on every committed rewrite. Model-visible means logged: anything that reaches a model request must be reconstructable from the log. The shared [row codec](src/chunk-rows.ts) losslessly converts event sequences to compact rows and back, preserves unrecognized events verbatim, and rejects malformed rows. Persistence backends decide whether to pack writes; bounded history transports can use the same rows while retaining the complete logical interval and exact decoding for consumers that need token boundaries.
+The package is built on event sourcing: a `Session` is an append-only log of typed `SessionEvent`s, and everything else — model history, transcripts, telemetry, titles, persistence — derives from that stream. The surface is a derived projection: an incremental manager validates append candidates, advances the ordered view from committed events, and tracks a `replaceGeneration` that bumps on every committed rewrite. Model-visible means logged: anything that reaches a model request must be reconstructable from the log. Each model attempt that reaches settlement commits one event: `assistant/message` carries the assembled model-visible message plus its compact timed stream, while `assistant/attempt` retains a failed, retried, cancelled, or stream-error attempt without adding model history. A hard process loss before settlement leaves no durable attempt stream.
 
 ### Request headers
 
@@ -92,7 +96,6 @@ The package is built on event sourcing: a `Session` is an append-only log of typ
 | [`src/surface.ts`](src/surface.ts) | Ordered surface projection, replacement validation, `deriveEventMessage` |
 | [`src/request-header.ts`](src/request-header.ts) | `request/header` folding and reconstruction |
 | [`dsh-util-values`](../../util/values/README.md) | Shared lossless JSON validation and detached snapshots |
-| [`src/chunk-rows.ts`](src/chunk-rows.ts) | Shared compact-row storage codec for persistence backends |
 | [`src/repair.ts`](src/repair.ts) | Cold repair of crash-orphaned logs |
 | [`src/invariant.ts`](src/invariant.ts) | Invariant companion: seq, turn/step enclosure, tool call/result pairing |
 
@@ -102,11 +105,11 @@ Every append uses the shared iterative `snapshotJsonValue()` pass, which reads,
 
 ### Derived history
 
-`deriveMessages()` caches each surface node's projection once and returns a fresh array per call over shared, deep-frozen messages; each of the three surface event types (`user/message`, `assistant/message`, `tool/result`) projects its own message kind — user content verbatim, the assembled assistant message with its provider and model, or a user-role tool result. A surface rewrite rebuilds the projection — there is no raw-log fallback, so the surface is the single source of derived history.
+`deriveMessages()` caches each surface node's projection once and returns a fresh array per call over shared, deep-frozen messages; each of the four surface event types (`system/message`, `user/message`, `assistant/message`, `tool/result`) projects its own message kind — the system-role prompt (an empty-content system node projects to no message), user content verbatim, the assembled assistant message with its provider and model, or a user-role tool result. Embedded Assistant streams and `assistant/attempt` events remain replay and diagnostic data only. A surface rewrite rebuilds the projection — there is no raw-log fallback, so the surface is the single source of derived history.
 
 ### The request header
 
-The loop logs a full canonical `request/header` snapshot (call config, adapter defaults, rendered system prompt, assembled tool schemas) at each loop-instance boundary and on change; `foldRequestHeader(events)` reconstructs it by selecting the latest snapshot, making every conversation request a pure function of the log. Route metadata (`request/context`) is separate logged state appended only when the provider, model, or capacity differs.
+The loop logs a full canonical `request/header` snapshot (call config, adapter defaults, assembled tool schemas — the rendered system prompt is a `system/message` surface node, not header state) at each loop-instance boundary and on change; `foldRequestHeader(events)` reconstructs it by selecting the latest snapshot, making every conversation request a pure function of the log. Route metadata (`request/context`) is separate logged state appended only when the provider, model, capacity, or `systemPromptUpdate` mode differs; it records the actual prepared call's mode after prompt and user admission, rather than supplying that admission decision.
 
 </details>
 
@@ -132,7 +135,7 @@ The package-level contract is enough for most consumers; read these when you nee
 
 #### What the model sees
 
-The model receives the complete messages from `user/message`, `assistant/message`, and `tool/result` surface entries verbatim — identities, roles, sources, and content blocks are the same values established at creation, and projections never mint identities. Direct prompts and injected context remain separate `user/message` events whose sources preserve their provenance. Chunks, boundaries, usage, and other log-only events add no message.
+The model receives the complete messages from `system/message`, `user/message`, `assistant/message`, and `tool/result` surface entries verbatim, the system prompt first — identities, roles, sources, and content blocks are the same values established at creation, and projections never mint identities. Direct prompts and injected context remain separate `user/message` events whose sources preserve their provenance. Embedded streams, `assistant/attempt`, boundaries, and other log-only facts add no message.
 
 #### Token effect
 
@@ -160,15 +163,15 @@ Append-only; newly visible content follows the reusable request prefix and does
 
 #### What the model sees
 
-The session reconstructs the system prompt, tool schemas, call config, and session prefix that the loop actually sent. Header events do not add a second copy to message history; the prefix is prepended outside `deriveMessages()`.
+The session reconstructs the tool schemas and call config that the loop actually sent; the system prompt is part of `deriveMessages()` as surface node 0 and, after an in-history update, as the latest system node. Header events add no message to history and hold no copy of the prompt.
 
 #### Token effect
 
-Zero duplicate tokens from logging. The reconstructed prefix, system text, and schemas still incur their normal per-request cost.
+Zero duplicate tokens from logging. The system nodes and schemas still incur their normal per-request cost.
 
 #### KV Cache effect
 
-Logging causes no invalidation, and exact reconstruction preserves request-prefix identity. A later header with changed prefix, prompt, or schemas may invalidate reuse from its first difference.
+Logging causes no invalidation, and exact reconstruction preserves request-prefix identity. A later header with changed config or schemas may invalidate reuse from its first difference; a prompt change that replaces surface node 0 invalidates reuse from the first token, while an in-history append keeps the prefix through the cached history reusable.
 
 ## Known Limitations and Deferred Work
 
@@ -177,8 +180,8 @@ Logging causes no invalidation, and exact reconstruction preserves request-prefi
 
 These limits define when the session store needs special care. They are current package constraints, not a task backlog.
 
-- **`fork()` cuts only at stable boundaries of live sessions** — the selected prefix must end outside an open turn and the source must be in the store; forking a persisted-but-unloaded session is excluded from the [fork API](../../../.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.md).
-- **`SESSION_FORMAT_VERSION` stays pinned at `0`** — pre-release, no broad compatibility implied: `Session` accepts only current seed shapes, a backend refuses any other version, and unknown event types refuse reconstruction unless marked `ignorable` in the envelope ([mechanism](../../../.agents/notes/implemented/architecture/2026-08-10-session-log-version-mechanism.md)).
+- **`fork()` cuts only at stable boundaries of live sessions** — the selected prefix must end outside an open turn and the source must be in the store; forking a persisted-but-unloaded session is excluded from the fork API.
+- **`SESSION_FORMAT_VERSION` names the current V3 logical representation** — the V3 reader rejects retired `header.system` and validates `system/message` payloads and protected-head rewrites. Historical headers and events belong to adjacent format packages; the V2→V3 edge converts supported history before constructing `Session`, and write open publishes only the V3 successor. Equal-version unknown events require the envelope's explicit `ignorable` marker, which does not promise safe structural migration ([mechanism](../../../.agents/notes/implemented/architecture/2026-08-31-released-session-format-migrations.md)).
 - **`TurnEndReasonMap` omits the ACP-named `refusal` / `max_turn_requests` variants** — producer-gated: they land when an adapter or the loop first emits them.
 - **No session tree beyond fork** — a pi-style entry tree over branched sessions is deferred unless a consumer needs more than boundary-based forking.
 

package/README.zh.md

@@ -9,7 +9,7 @@ kind: "package-reference"
 
 ## 概述
 
-`dsh-session` 提供仅追加的会话日志，记录 agent（智能体）的完整交互历史——每个模型可见事实都流经的单一真源。LLM（大语言模型）消息历史由日志*派生*（`deriveMessages()`），从不另行存储，因此回放就是对同一批事件重新派生，压缩（compaction）也可以遮蔽较旧的表层条目而不删除历史。该包还提供内存存储（`ctx.sessions`）、插件通过声明合并扩展的类型化 `SessionEvent` 词汇，以及为产生消息的事件排序的 surface 层。持久化刻意是独立关注点：后端订阅 `session/event` 并在 `session/flush` 时刷新。作为任何 agent 会话的基础时请选择本包；它本身不运行模型调用。
+`dsh-session` 在仅追加的会话日志中记录每个模型可见事实，并从该记录派生模型历史。消费方可以检查、回放、派生和刷新会话，同时保留历史事件；压缩会在活跃对话中隐藏被取代的条目，但不会删除它们。除非添加持久化后端，否则会话仅保留在内存中；持久性检查点会等待配置的后端。agent 需要可重建的会话记录时请选择本包；它本身不调用模型。
 
 ## 目录
 
@@ -47,7 +47,11 @@ session.append('user/message', { role: 'user', content: [{ type: 'text', text: '
 session.deriveMessages()         // the derived model history
 ```
 
-表层事件（`user/message`、`assistant/message`、`tool/result`）必须声明如何进入有序 surface；原始分片、边界与其他仅日志事件从不产生消息。
+表层事件（`system/message`、`user/message`、`assistant/message`、`tool/result`）在类型化事件与追加输入中都必须带有 `surfaceOp`。替换操作仅接受 `{ op: 'replace', startSeq, endSeq }`，端点为包含边界的 `SessionSeq`，按当前 surface 顺序解释。Assistant message 嵌入其精确紧凑 provider stream，并禁止 `sourceEventSeqs`。已知仅日志事件禁止这两个元数据字段，且从不产生消息。
+
+追加、seed/restore 与事件 adoption/snapshot 会拒绝任何 `header.system` 及恰好为空的可选请求头字段（`tools: []`、`adapterDefaults: {}`），而不规范化输入。工具结果的 `data.error` 仅在 `message.content[0].isError === true` 时允许存在；失败标识仍是可选的。被拒绝的追加不会改变日志、派生状态或事件流。Adoption 校验事件局部元数据，但不校验所引用的历史或替换端点是否属于 surface。
+
+`system/message` 承载渲染后的系统提示词：第一条是 surface 第 0 号节点，准入依据已准备调用的能力，不具备能力的路由将非空渲染文本归并到首个系统节点，延续中的 `in-history` 序列则在缓存历史之后追加；空系统节点不投影为消息，因此清除提示词必须为所有生效的系统节点记录空内容替换，而非仅替换最新节点；当第 0 号节点是 `system/message` 时，surface 折叠拒绝覆盖它的替换，除非替换事件本身是恰好覆盖该节点的 `system/message`，而后续系统节点不受保护，压缩范围可以遮蔽它们（[决策](../../../.agents/notes/implemented/architecture/2026-09-02-system-prompt-as-surface-node.zh.md)）。
 
 ### 读取日志
 
@@ -77,7 +81,7 @@ session.deriveMessages()         // the derived model history
 
 ### 设计理念
 
-该包建立在事件溯源之上：`Session` 是类型化 `SessionEvent` 的仅追加日志，其他一切——模型历史、transcript、遥测、标题、持久化——都从这条流派生。surface 是派生投影：一个增量管理器校验追加候选、根据已提交事件推进有序视图，并跟踪每次已提交重写都会递增的 `replaceGeneration`。模型可见即已记录：任何到达模型请求的内容都必须能从日志重建。共享的[行编解码器](src/chunk-rows.ts)在事件序列与紧凑行之间无损转换，逐字保留无法识别的事件，并拒绝形态错误的行。持久化后端决定是否打包写入；有界历史传输可以使用同一种行，同时保留完整逻辑区间，并为需要 token 边界的消费方提供精确解码。
+该包建立在事件溯源之上：`Session` 是类型化 `SessionEvent` 的仅追加日志，其他一切——模型历史、transcript、遥测、标题、持久化——都从这条流派生。surface 是派生投影：一个增量管理器校验追加候选、根据已提交事件推进有序视图，并跟踪每次已提交重写都会递增的 `replaceGeneration`。模型可见即已记录：任何到达模型请求的内容都必须能从日志重建。每个到达 settlement 的模型 attempt 都会提交一个事件：`assistant/message` 携带组装后的模型可见 message 及其紧凑带时间 stream，`assistant/attempt` 则保留失败、重试、取消或 stream error attempt，且不添加模型历史。如果进程在 settlement 前硬中断，则不会留下持久 attempt stream。
 
 ### 请求 header
 
@@ -92,7 +96,6 @@ session.deriveMessages()         // the derived model history
 | [`src/surface.ts`](src/surface.ts) | 有序 surface 投影、替换校验、`deriveEventMessage` |
 | [`src/request-header.ts`](src/request-header.ts) | `request/header` 折叠与重建 |
 | [`dsh-util-values`](../../util/values/README.zh.md) | 共享无损 JSON 校验与分离式快照 |
-| [`src/chunk-rows.ts`](src/chunk-rows.ts) | 供持久化后端使用的共享紧凑行存储编解码器 |
 | [`src/repair.ts`](src/repair.ts) | 崩溃遗留日志的冷修复 |
 | [`src/invariant.ts`](src/invariant.ts) | 不变式配套：序号、轮次／步骤闭合、工具调用／结果配对 |
 
@@ -102,11 +105,11 @@ session.deriveMessages()         // the derived model history
 
 ### 派生历史
 
-`deriveMessages()` 把每个 surface 节点的投影缓存一次，每次调用都返回共享、深度冻结消息之上的新数组；三种 surface 事件类型（`user/message`、`assistant/message`、`tool/result`）各自投影自己的消息种类——user 内容原样、带提供方与模型的组装 assistant 消息，或 user 角色的工具结果。surface 重写会重建投影——不存在原始日志回退，因此 surface 是派生历史的唯一来源。
+`deriveMessages()` 把每个 surface 节点的投影缓存一次，每次调用都返回共享、深度冻结消息之上的新数组；四种 surface 事件类型（`system/message`、`user/message`、`assistant/message`、`tool/result`）各自投影自己的消息种类——system 角色的提示词（空内容的系统节点投影为无消息）、user 内容原样、带提供方与模型的组装 assistant 消息，或 user 角色的工具结果。嵌入式 Assistant stream 与 `assistant/attempt` 事件只保留重放和诊断数据。surface 重写会重建投影——不存在原始日志回退，因此 surface 是派生历史的唯一来源。
 
 ### 请求头
 
-循环在每个循环实例边界及变更时记录完整规范 `request/header` 快照（调用配置、适配器默认值、渲染后的系统提示词、组装后的工具 schema）；`foldRequestHeader(events)` 通过选择最新快照来重建它，使每个对话请求都成为日志的纯函数。路由元数据（`request/context`）是独立的已记录状态，仅在提供方、模型或容量变化时追加。
+循环在每个循环实例边界及变更时记录完整规范 `request/header` 快照（调用配置、适配器默认值、组装后的工具 schema——渲染后的系统提示词是 `system/message` surface 节点，不是 header 状态）；`foldRequestHeader(events)` 通过选择最新快照来重建它，使每个对话请求都成为日志的纯函数。路由元数据（`request/context`）是独立的已记录状态，仅在提供方、模型、容量或 `systemPromptUpdate` 模式变化时追加；它在提示词与用户消息准入之后记录实际已准备调用的模式，而非提供准入决策。
 
 </details>
 
@@ -132,7 +135,7 @@ session.deriveMessages()         // the derived model history
 
 #### 模型看到什么
 
-模型会原样接收 `user/message`、`assistant/message` 与 `tool/result` surface 条目中的完整消息——标识、角色、来源与内容块都与创建时确定的值相同，投影从不生成标识。直接提示词与注入上下文仍是彼此独立的 `user/message` 事件，各事件的来源会保留其出处。分片、边界、用量与其他仅日志事件不会添加消息。
+模型会原样接收 `system/message`、`user/message`、`assistant/message` 与 `tool/result` surface 条目中的完整消息，系统提示词在先——标识、角色、来源与内容块都与创建时确定的值相同，投影从不生成标识。直接提示词与注入上下文仍是彼此独立的 `user/message` 事件，各事件的来源会保留其出处。嵌入式 stream、`assistant/attempt`、边界与其他仅日志事实不会添加消息。
 
 #### Token 影响
 
@@ -160,15 +163,15 @@ session.deriveMessages()         // the derived model history
 
 #### 模型看到什么
 
-会话会重建循环实际发送的系统提示词、工具 schema、调用配置与会话前缀。请求头事件不会向消息历史加入第二份副本；前缀在 `deriveMessages()` 外部前置。
+会话会重建循环实际发送的工具 schema 与调用配置；系统提示词作为 surface 第 0 号节点、并在历史内更新之后作为最新的系统节点，属于 `deriveMessages()` 的一部分。请求头事件不向历史加入任何消息，也不持有提示词的副本。
 
 #### Token 影响
 
-日志记录不产生重复 token。重建的前缀、系统文本与 schema 仍会产生正常的逐请求开销。
+日志记录不产生重复 token。各系统节点与 schema 仍会产生正常的逐请求开销。
 
 #### KV Cache 影响
 
-记录日志不会导致失效，精确重建会保持请求前缀一致。后续请求头若更改前缀、提示词或 schema，可能从第一处差异开始使复用失效。
+记录日志不会导致失效，精确重建会保持请求前缀一致。后续请求头若更改配置或 schema，可能从第一处差异开始使复用失效；替换 surface 第 0 号节点的提示词变更会从第一个 token 起使复用失效，而历史内追加则保持直到已缓存历史末尾的前缀可复用。
 
 ## 已知限制与延期工作
 
@@ -177,8 +180,8 @@ session.deriveMessages()         // the derived model history
 
 这些限制说明会话存储何时需要特别留意。它们是当前包约束，不是任务积压。
 
-- **`fork()` 仅在实时会话的稳定边界处切分**：所选前缀结束时不得有开放轮次，且源会话必须位于存储中；[fork API](../../../.agents/notes/implemented/feature/2026-06-30-session-store-fork-api.zh.md) 不支持对已持久化但未加载的会话进行 fork。
-- **`SESSION_FORMAT_VERSION` 固定为 `0`**：预发布阶段不承诺广泛兼容性；`Session` 只接受当前 seed 形状，后端拒绝任何其他版本，不认识的事件类型也会拒绝重建，除非信封带 `ignorable` 标记（[机制](../../../.agents/notes/implemented/architecture/2026-08-10-session-log-version-mechanism.zh.md)）。
+- **`fork()` 仅在实时会话的稳定边界处切分**：所选前缀结束时不得有开放轮次，且源会话必须位于存储中；fork API 不支持对已持久化但未加载的会话进行 fork。
+- **`SESSION_FORMAT_VERSION` 命名当前 V3 逻辑表示**——V3 读取器拒绝已退役的 `header.system`，并校验 `system/message` 载荷与受保护头节点的重写。历史 header 与事件归相邻格式包所有；V2→V3 迁移边在构造 `Session` 前转换受支持的历史，写打开只发布 V3 后继代际。同版本未知事件要求信封显式带有 `ignorable` 标记，但这不保证结构迁移的安全性（[机制](../../../.agents/notes/implemented/architecture/2026-08-31-released-session-format-migrations.zh.md)）。
 - **`TurnEndReasonMap` 不含 ACP（Agent Client Protocol）命名的 `refusal`／`max_turn_requests` 变体**：受生产方约束；只有当适配器或循环首次产生这些变体时才加入。
 - **fork 之外没有会话树**：基于分支会话的 pi 风格条目树被推迟，除非消费方需要超越基于边界的 forking 的能力。