docs: replace vague provenance prose with recorded facts

.agents/notes/implemented/architecture/2026-06-18-session-surface.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 .agents/notes/implemented/architecture/2026-06-18-session-surface.md
-2026-06-18-session-surface.md: 1bb3baac5e9bea3f657cbd3be2623223bda78ce2
-2026-06-18-session-surface.zh.md: b967ad467e58ed4f4b4009fe4746f1b8b6b22f25
+2026-06-18-session-surface.md: 3a033960e7cbf577c69c1df9940e064ef34163d3
+2026-06-18-session-surface.zh.md: 5c1348595589e8b69b0fe4a679f2536a83ef3142

.agents/notes/implemented/architecture/2026-06-18-session-surface.md

@@ -6,7 +6,7 @@ English | [中文](2026-06-18-session-surface.zh.md)
 
 ## Problem
 
-The event log is authoritative, but history manipulation had no durable shared mechanism. Plugins such as compaction would otherwise rewrite derived requests through order-sensitive listeners, leave no provenance, and require repeated changes to `deriveMessages()`.
+The event log is authoritative, but history manipulation had no durable shared mechanism. Without one, plugins such as compaction would rewrite derived requests through order-sensitive listeners without recording which events each replacement used. Every new history manipulation would also require changes to `deriveMessages()`.
 
 ## Decision
 
@@ -16,7 +16,7 @@ Add a **surface** — a derived, cached order of event sequences (the subset of
 
 Every `SessionEvent` gains two optional fields (structural metadata, like `seq`/`time`):
 
-- **`sourceEventSeqs?: number[]`** — seq numbers of events that are provenance sources (e.g., the `assistant/chunk` seqs that built an `assistant/message`, or the surface nodes shadowed by a compaction marker). A present `[]` is valid only on `assistant/message` and records a known empty provider stream; omission there means legacy or otherwise unrecorded provenance. Other surface events require a non-empty list when the field is present. Provenance is a core design principle; without it, the replace-range operation cannot be validated on replay.
+- **`sourceEventSeqs?: number[]`** — seq numbers of earlier events cited as sources (e.g., the `assistant/chunk` seqs that built an `assistant/message`, or the surface nodes shadowed by a compaction marker). A present `[]` is valid only on `assistant/message` and records a known empty provider stream; omission there means a legacy or foreign event did not record the stream. Other surface events require a non-empty list when the field is present. Without these cited seqs, replay cannot validate that a replace-range operation names every event it removed.
 - **`surfaceOp?: SurfaceOp`** — how this event entered the surface. Absent for non-surface events.
 
 ### SurfaceOp: two operations
@@ -49,7 +49,7 @@ The `repair.ts` module synthesizes `tool/result` closers for orphaned tool calls
 
 ### Invariants
 
-`Session` validates `sourceEventSeqs` and `surfaceOp` at the always-on seed/append boundary: only `assistant/message` may use an empty provenance list; references are unique, earlier, and known; replacement endpoints exist in surface order; and provenance covers every shadowed node. These are single-record acceptance and storage-projection rules, not optional invariant-service contributions.
+`Session` validates `sourceEventSeqs` and `surfaceOp` at the always-on seed/append boundary: only `assistant/message` may use an empty source-event list; references are unique, earlier, and known; replacement endpoints exist in surface order; and `sourceEventSeqs` covers every shadowed node. These are single-record acceptance and storage-projection rules, not optional invariant-service contributions.
 
 Every surface-eligible event must carry `surfaceOp` or it would disappear from derived history. Typed `append` overloads enforce this for literal event types; runtime checks in `append` and the seed constructor cover widened unions and loaded logs. Invalid seeds are rejected rather than upgraded under the pre-release format policy.
 
@@ -63,11 +63,11 @@ Every surface-eligible event must carry `surfaceOp` or it would disappear from d
 ## Consequences
 
 - **`packages/core/session`**: `surface.ts` (`SurfaceManager`) maintains one ordered seq array for candidate acceptance and live projection; `SessionSurface` is its readonly public view. `SurfaceOp`/`SurfaceIntent` and the top-level session-event fields record how entries join it. `append()` requires a `SurfaceIntent` for surface events, `deriveMessages()` walks the surface as the sole derivation path, and `repair.ts` emits surface-aware closers. The seed constructor rejects a surface-eligible seed event missing its `surfaceOp` marker (see § Invariants).
-- **`packages/core/agent-loop`**: All surface-capable appends pass surface opts. Chunk seqs are collected for `assistant/message` provenance; `tool/call` seqs are captured for `tool/result` provenance.
+- **`packages/core/agent-loop`**: All surface-capable appends pass surface opts. Each `assistant/message` cites its chunk seqs; each `tool/result` cites its `tool/call` seq.
 - **`packages/session/session-persistence-sqlite`**: Two new nullable TEXT columns (`source_event_seqs`, `surface_op`) on the `events` table; `SCHEMA_VERSION` bumped (bump-and-reject, no migration).
 - **`packages/session/session-persistence-jsonl`**: No changes required.
 - **`packages/session/session-persistence`**: Abstract interface unchanged.
 
 The surface is the foundation for future history manipulation. A compaction or tool-result-prune plugin appends one of the existing message-producing event types (a `user/message` carrying the summary, say) with `surfaceOp: { op: 'replace', start, end }` and `sourceEventSeqs` covering the shadowed entries — the new event takes the range's place on the surface while the plugin's own trace events (e.g. `compaction/start`, `compaction/end`) stay off it. Replay preserves the decision deterministically.
 
-A `tool/result` replacement may rewrite exactly one current `tool/result` and must preserve every data field except `content`. Session acceptance enforces this rule together with positional range and provenance validation, independent of optional diagnostic plugins.
+A `tool/result` replacement may rewrite exactly one current `tool/result` and must preserve every data field except `content`. Session acceptance enforces this rule together with positional range and cited source-event validation, independent of optional diagnostic plugins.

.agents/notes/implemented/architecture/2026-06-18-session-surface.zh.md

@@ -6,7 +6,7 @@ Status: implemented
 
 ## 问题
 
-事件日志是权威数据源，但历史操纵此前没有持久化的共享机制。如果没有这样的机制，上下文压缩（context compaction）等插件只能通过顺序敏感的监听器改写派生请求，不留溯源信息，且每次新增操纵都要反复修改 `deriveMessages()`。
+事件日志是权威数据源，但历史操纵此前没有持久化的共享机制。如果没有这样的机制，上下文压缩（context compaction）等插件会通过顺序敏感的监听器改写派生请求，却不记录每次替换使用了哪些事件。每次新增历史操纵时，还必须修改 `deriveMessages()`。
 
 ## 决策
 
@@ -16,7 +16,7 @@ Status: implemented
 
 每个 `SessionEvent` 获得两个可选字段（结构性元数据，与 `seq`/`time` 同级）：
 
-- **`sourceEventSeqs?: number[]`**：作为溯源来源的事件 seq 编号（例如构成 `assistant/message` 的各 `assistant/chunk` 的 seq，或被压缩标记遮蔽的 surface 节点）。出现的 `[]` 只在 `assistant/message` 上有效，表示已知为空的提供方流；在该事件上省略字段表示旧数据或未记录的溯源。其他 surface 事件一旦出现此字段，就必须是非空列表。溯源是核心设计原则；没有它，replace-range 操作在回放时无法被验证。
+- **`sourceEventSeqs?: number[]`**：被引用为数据来源的早期事件 seq 编号（例如构成 `assistant/message` 的各 `assistant/chunk` 的 seq，或被压缩标记遮蔽的 surface 节点）。出现的 `[]` 只在 `assistant/message` 上有效，表示已知为空的提供方流；在该事件上省略字段，表示某个旧事件或外部事件没有记录该流。其他 surface 事件一旦出现此字段，就必须是非空列表。如果没有这些引用的 seq，回放就无法验证 replace-range 操作是否列出了它移除的每个事件。
 - **`surfaceOp?: SurfaceOp`**：该事件如何进入 surface。非 surface 事件不携带此字段。
 
 ### SurfaceOp：两种操作
@@ -49,7 +49,7 @@ export type SurfaceOp =
 
 ### 不变式
 
-`Session` 在始终启用的 seed/append 边界校验 `sourceEventSeqs` 与 `surfaceOp`：只有 `assistant/message` 可以使用空的溯源列表；引用必须唯一、更早且已知；替换端点必须存在于 surface 顺序中；溯源必须覆盖每个被遮蔽的节点。这些是单记录接纳与存储投影规则，不是由可选的不变式服务提供的规则。
+`Session` 在始终启用的 seed/append 边界校验 `sourceEventSeqs` 与 `surfaceOp`：只有 `assistant/message` 可以使用空的源事件列表；引用必须唯一、更早且已知；替换端点必须存在于 surface 顺序中；`sourceEventSeqs` 必须覆盖每个被遮蔽的节点。这些是单记录接纳与存储投影规则，不是由可选的不变式服务提供的规则。
 
 每个可进入 surface 的事件都必须携带 `surfaceOp`，否则它将从派生历史中消失。类型化的 `append` 重载对字面事件类型强制执行此规则；`append` 和种子构造函数中的运行时检查覆盖宽化联合类型和加载的日志。按照预发布格式策略，无效的种子被拒绝而非升级。
 
@@ -63,11 +63,11 @@ export type SurfaceOp =
 ## 后果
 
 - **`packages/core/session`**：`surface.ts`（`SurfaceManager`）维护一个用于候选接纳和实时投影的有序 seq 数组；`SessionSurface` 是其只读公共视图。`SurfaceOp`/`SurfaceIntent` 与顶层会话事件字段记录条目如何加入它。`append()` 要求 surface 事件携带 `SurfaceIntent`，`deriveMessages()` 以遍历 surface 作为唯一派生路径，`repair.ts` 则发出 surface 感知的闭合事件。种子构造函数拒绝缺少 `surfaceOp` 标记的可进入 surface 的种子事件（见「不变式」一节）。
-- **`packages/core/agent-loop`**：所有涉及 surface 事件的追加操作都传入 surface 选项。收集分片 seq 用于 `assistant/message` 溯源；捕获 `tool/call` seq 用于 `tool/result` 溯源。
+- **`packages/core/agent-loop`**：所有涉及 surface 事件的追加操作都传入 surface 选项。每个 `assistant/message` 都引用产生它的分片 seq；每个 `tool/result` 都引用它的 `tool/call` seq。
 - **`packages/session/session-persistence-sqlite`**：`events` 表新增两个可空 TEXT 列（`source_event_seqs`、`surface_op`）；`SCHEMA_VERSION` 递增（bump-and-reject，无迁移）。
 - **`packages/session/session-persistence-jsonl`**：无需改动。
 - **`packages/session/session-persistence`**：抽象接口不变。
 
 Surface 是未来历史操纵的基础。压缩或 tool-result-prune 插件追加一个既有的消息产出事件类型（例如一条携带摘要的 `user/message`），附带 `surfaceOp: { op: 'replace', start, end }` 和覆盖被遮蔽条目的 `sourceEventSeqs`——新事件在 surface 上取代该范围的位置，而插件自身的 trace 事件（如 `compaction/start`、`compaction/end`）不进入 surface。回放以确定性方式保留该决策。
 
-一次 `tool/result` 替换只能改写当前的一个 `tool/result`，并且必须保留除 `content` 以外的每个数据字段。Session 接纳会与位置范围和溯源校验一起强制这条规则，不依赖可选的诊断插件。
+一次 `tool/result` 替换只能改写当前的一个 `tool/result`，并且必须保留除 `content` 以外的每个数据字段。Session 接纳会与位置范围和引用的源事件校验一起强制这条规则，不依赖可选的诊断插件。

.agents/notes/implemented/architecture/2026-06-21-bounded-llm-request-recovery.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 .agents/notes/implemented/architecture/2026-06-21-bounded-llm-request-recovery.md
-2026-06-21-bounded-llm-request-recovery.md: 587f2d26eee922e91c8797018964f983890eb8ec
-2026-06-21-bounded-llm-request-recovery.zh.md: dfc5e3ab67b62042c37f05199ed39bf47ded0c5d
+2026-06-21-bounded-llm-request-recovery.md: 41f02f977a5fb628d57c162c64f8aaa1c2e6b1c0
+2026-06-21-bounded-llm-request-recovery.zh.md: 3e95865d73e75b8f7ae227ebabe322a21c2b83ae

.agents/notes/implemented/architecture/2026-06-21-bounded-llm-request-recovery.md

@@ -115,7 +115,7 @@ If recovery is exhausted, the final failure is stored once on `turn/end.reason`
 - HMR-during-backoff tests prove disposal unregisters the listener, aborts and awaits its captured callbacks, emits no retry decision after disposal, and leaves no timer or promise alive.
 - Pure unit tests cover transient-code selection, exponential backoff and jitter bounds, valid and over-cap `Retry-After`, exhausted budgets, deterministic timer/random seams, and abort during backoff.
 - Real agent-loop tests cover failure before chunks, partial chunks then failure, thrown and in-band failures, retry to success in a new turn, exhaustion to structured `turn/end.reason`, and composition with `dsh-compact-basic` context-overflow recovery.
-- The partial-chunk integration test proves failed chunks remain attributed to the failed step, no assistant message or tool side effect is committed for that step, and the successful retry has distinct provenance.
+- The partial-chunk integration test proves failed chunks remain attributed to the failed step, no assistant message or tool side effect is committed for that step, and the successful retry records its own chunk seqs and provider/model route.
 - The plugin-owned `llm/retry` event is non-surface, survives JSONL and SQLite round trips, is ignored by message derivation, and drives TUI and Web retraction plus scheduled-retry rendering. Client tests cover complete wire validation, clock-independent countdown, cancellation versus completed retry labels, and trajectory attribution; keyless UI snapshots cover Web scheduling and success, a real Web composition test covers partial transport failure through recovery, and ACP automation snapshots confirm that a discarded attempt stays off the wire while the recovered reply is emitted.
 - Idle-watchdog tests prove the stable signal is rearmed only while `next()` is outstanding, disarmed during consumer think time and in `finally`, and classified separately from a total-call deadline and an earlier caller abort; adapter tests prove the signal stops the underlying request rather than merely detaching it.
 - Direct `ctx.llm.stream()` callers remain single-attempt and receive the same structured failure facts.

.agents/notes/implemented/architecture/2026-06-21-bounded-llm-request-recovery.zh.md

@@ -115,7 +115,7 @@ agent-spine 演示组合包加载该插件，因此共享的 stdio/TUI、一次
 - 退避期间执行 HMR 的测试证明：dispose 过程会注销监听器、中止并等待其捕获的回调，dispose 后不发出重试决策，也不留下存活的定时器或 promise。
 - 纯单元测试覆盖暂时性 code 选择、指数退避和抖动边界、有效及超出上限的 `Retry-After`、耗尽的预算、确定性定时器／随机数 seam，以及退避期间中止。
 - 真实 agent-loop 测试覆盖分片前失败、部分分片后失败、抛出及带内失败、在新轮次中重试至成功、耗尽后写入结构化 `turn/end.reason`，以及与 `dsh-compact-basic` 上下文溢出恢复的组合。
-- 部分分片集成测试证明：失败分片仍归属于失败步骤，该步骤不会提交 assistant 消息或工具副作用，成功的重试具有不同的来源信息。
+- 部分分片集成测试证明：失败分片仍归属于失败步骤，该步骤不会提交 assistant 消息或工具副作用，成功的重试会记录自己的分片 seq 和提供方／模型路由。
 - 插件拥有的不进入表层的 `llm/retry` 事件可在 JSONL 和 SQLite 往返后保留，被消息派生忽略，并驱动 TUI 和 Web 撤回及计划重试渲染。客户端测试覆盖完整的 wire 验证、独立于时钟的倒计时、已取消与已完成重试标签的区别以及轨迹归属；无密钥 UI 快照覆盖 Web 的调度与成功，真实 Web 组合测试覆盖部分传输失败直至恢复，ACP 自动化快照确认，被丢弃的尝试不会通过协议发出，而恢复后的回复会正常发出。
 - 空闲看门狗测试证明：只有 `next()` 尚未完成时才会重新布防稳定信号；在消费方思考期间及 `finally` 中会解除布防；它与总调用 deadline 以及更早发生的调用方中止分开分类。适配器测试证明该信号会终止底层请求，而不只是与其脱离。
 - `ctx.llm.stream()` 的直接调用方仍只尝试一次，并收到相同的结构化失败事实。

.agents/notes/implemented/architecture/2026-07-05-prompt-variables-and-tool-guidance-ownership.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 .agents/notes/implemented/architecture/2026-07-05-prompt-variables-and-tool-guidance-ownership.md
-2026-07-05-prompt-variables-and-tool-guidance-ownership.md: a3b5021daf323971308760bde4f97651db8edbda
-2026-07-05-prompt-variables-and-tool-guidance-ownership.zh.md: 0b84f188e6637d8b1b42ed2e6df3cd7e4061d75d
+2026-07-05-prompt-variables-and-tool-guidance-ownership.md: 199153741a5677e9f7b9bc6510f7a77b9791343e
+2026-07-05-prompt-variables-and-tool-guidance-ownership.zh.md: 988c128382fd8ff5a8a4742486860d931013ac80

.agents/notes/implemented/architecture/2026-07-05-prompt-variables-and-tool-guidance-ownership.md

@@ -18,7 +18,7 @@ The assembled system prompt had four defects, all of one family: facts the harne
 
 ## Decision
 
-**One principle: every fact in the prompt has exactly one owner.** The model name and workspace are config/session facts → the harness exposes them as variables and the persona references them. Per-tool semantics and when-to-use → the tool's `description`. Cross-call habits a description cannot carry → the tool package's prompt section. Harness provenance → the static `harness:identity` section. Deployment role and behavior → the deployment's persona.
+**One principle: every fact in the prompt has exactly one owner.** The model name and workspace are config/session facts → the harness exposes them as variables and the persona references them. Per-tool semantics and when-to-use → the tool's `description`. Cross-call habits a description cannot carry → the tool package's prompt section. The product name and SDK identity line → the static `harness:identity` section. Deployment role and behavior → the deployment's persona.
 
 ### Assemble context
 

.agents/notes/implemented/architecture/2026-07-05-prompt-variables-and-tool-guidance-ownership.zh.md

@@ -18,7 +18,7 @@ Status: implemented
 
 ## 决策
 
-**一条原则：提示词中的每个事实恰好有一个归属方。** 模型名称和工作区是配置/会话事实 → harness 将它们暴露为变量，persona 引用它们。每个工具的语义和何时使用 → 工具的 `description`。description 无法承载的跨调用习惯 → 包的提示词 section。harness 来源标识 → 静态的 `harness:identity` section。部署角色与行为 → 部署的 persona。
+**一条原则：提示词中的每个事实恰好有一个归属方。** 模型名称和工作区是配置/会话事实 → harness 将它们暴露为变量，persona 引用它们。每个工具的语义和何时使用 → 工具的 `description`。description 无法承载的跨调用习惯 → 包的提示词 section。产品名称和 SDK 身份说明 → 静态的 `harness:identity` section。部署角色与行为 → 部署的 persona。
 
 ### 组装上下文
 

.agents/notes/implemented/architecture/2026-07-05-reconstructable-requests.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 .agents/notes/implemented/architecture/2026-07-05-reconstructable-requests.md
-2026-07-05-reconstructable-requests.md: ebca9b99cad791159302da9c2bbce9f4df147aab
-2026-07-05-reconstructable-requests.zh.md: e7ed83bbc553e86d53d88e442b0fd8992973c9e7
+2026-07-05-reconstructable-requests.md: 3676e4bc7f6d0272f382b13e178f04b16f18f992
+2026-07-05-reconstructable-requests.zh.md: bcae6b941fb0110aa4e5011685365a5edf7512d3

.agents/notes/implemented/architecture/2026-07-05-reconstructable-requests.md

@@ -30,9 +30,9 @@ Each proposed step first claims its inbox batch and runs `agent/pre-step`. Rejec
 
 **Enforcement.** The `dsh-agent-loop/invariant` companion registers with `ctx.invariants` and, when selected, independently rebuilds each loop request through a fresh `Session`, so the live cache cannot vouch for itself, then compares messages and folded header fields at `llm/stream`. The loop records the exact frozen request through `markAgentLoopRequest()` in `dsh-llm`; the process-local identity lets the companion and other request observers recognize conversation work, while direct one-shots remain excluded regardless of their frozen shape or session id. Correctness depends on sequence-bounded reconstruction rather than listener order. A with-key e2e requires positive cache-read tokens after the first request; per-step usage is the production signal, and a header change or compaction appears as a cache-read drop on the next step.
 
-### The MiniCode shape: adopted, with the provenance arrow inverted
+### The MiniCode shape: adopted, with the event log as the source
 
-Like MiniCode, the conversation advances append-only and resets only when model-visible state changes. Unlike MiniCode, the event log remains the source of truth because it also owns persistence, recovery, boundaries, tool pairing, and provenance. `Session` caches message and header folds derived from that log, making every request independently checkable.
+Like MiniCode, the conversation advances append-only and resets only when model-visible state changes. Unlike MiniCode, the event log remains the source of truth because it also owns persistence, recovery, boundaries, tool pairing, and links from derived events to their inputs. `Session` caches message and header folds derived from that log, making every request independently checkable.
 
 ## Alternatives considered
 

.agents/notes/implemented/architecture/2026-07-05-reconstructable-requests.zh.md

@@ -30,9 +30,9 @@ Status: implemented
 
 **强制执行。** `dsh-agent-loop/invariant` 配套插件向 `ctx.invariants` 注册，并在被选用时通过一个全新的 `Session` 独立重建每个循环请求，使活跃缓存无法为自身背书，然后在 `llm/stream` 处比较消息和折叠后的 header 字段。循环通过 `dsh-llm` 的 `markAgentLoopRequest()` 记录精确的冻结请求；这一进程内标识让配套插件和其他请求观察者识别对话工作，而直接的一次性调用无论其冻结形状或会话 id 如何都保持排除。正确性依赖于序列有界的重建，而非监听器顺序。带密钥的 e2e 要求首次请求之后有正值的 cache-read token；逐步骤用量是生产信号，header 变更或压缩表现为下一步骤的 cache-read 下降。
 
-### MiniCode 形态：采纳，但溯源箭头反转
+### MiniCode 形态：采纳，以事件日志为真源
 
-与 MiniCode 相同，对话仅追加推进，仅在模型可见状态变更时重置。与 MiniCode 不同，事件日志仍是真源，因为它同时拥有持久化、恢复、边界、工具配对和溯源。`Session` 缓存从日志推导的消息和 header 折叠结果，使每个请求都可独立检查。
+与 MiniCode 相同，对话仅追加推进，仅在模型可见状态变更时重置。与 MiniCode 不同，事件日志仍是真源，因为它同时拥有持久化、恢复、边界、工具配对，并记录派生事件到其输入事件的关联。`Session` 缓存从日志推导的消息和 header 折叠结果，使每个请求都可独立检查。
 
 ## 曾考虑的替代方案
 

.agents/notes/implemented/architecture/2026-07-10-after-call-compaction-pressure-and-overflow-recovery.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 .agents/notes/implemented/architecture/2026-07-10-after-call-compaction-pressure-and-overflow-recovery.md
-2026-07-10-after-call-compaction-pressure-and-overflow-recovery.md: 6fbd5e2c9d57da3f25c72c652ca50eb45b84323c
-2026-07-10-after-call-compaction-pressure-and-overflow-recovery.zh.md: 8a480999038cf892842290e3ea3902683d87e431
+2026-07-10-after-call-compaction-pressure-and-overflow-recovery.md: 6f18d5da014f35b47c40c541ca75ca5ab81499b2
+2026-07-10-after-call-compaction-pressure-and-overflow-recovery.zh.md: 7b6ce75b145e96e58a49331e0390f1d3f82200e4

.agents/notes/implemented/architecture/2026-07-10-after-call-compaction-pressure-and-overflow-recovery.md

@@ -32,7 +32,7 @@ If cancellation lands after assistant tool calls are durable but before all call
 
 `CompactService.compactIfNeeded(agent, trigger, signal)` accepts `trigger: 'pressure' | 'context-overflow'`. The interface gains no estimation methods or token types; `ctx.tokenMeter` remains the reusable accounting owner.
 
-For `pressure`, compact-basic resolves the durable provider/model target's adapter-owned capacity and exact-target policy, then applies the resulting threshold and retained-tail budgets to one unified `ctx.tokenMeter.measure()` result. Below pressure it returns without pruning. Once pressure qualifies, optional `ctx.toolResultPrune` rewrites oversized current results and compact-basic remeasures through the same meter; safe pressure skips the model call, while remaining pressure selects and summarizes from the pruned surface. The same singleton meter owns range pricing, provenance, shadowed token counts, and non-shrinking-summary rejection. Common defaults remain threshold ratio `0.8`, retained-history ratio `0.16`, summarization provider/model `''`, `maxTokens: 8192`, `compactionRetries: 1`, and `auto: true`; optional `modelPolicies` entries override them for an exact provider/model pair.
+For `pressure`, compact-basic resolves the durable provider/model target's adapter-owned capacity and exact-target policy, then applies the resulting threshold and retained-tail budgets to one unified `ctx.tokenMeter.measure()` result. Below pressure it returns without pruning. Once pressure qualifies, optional `ctx.toolResultPrune` rewrites oversized current results and compact-basic remeasures through the same meter; safe pressure skips the model call, while remaining pressure selects and summarizes from the pruned surface. The same singleton meter owns range pricing, cited source-event accounting, shadowed token counts, and non-shrinking-summary rejection. Common defaults remain threshold ratio `0.8`, retained-history ratio `0.16`, summarization provider/model `''`, `maxTokens: 8192`, `compactionRetries: 1`, and `auto: true`; optional `modelPolicies` entries override them for an exact provider/model pair.
 
 For canonical overflow, compact-basic requires no capacity metadata and bypasses scalar pressure and the normal retained-token budget. It prunes first, then chooses the maximal tool-balanced head range while leaving the newest indivisible unit and attempts one shrinking summary compaction under the same signal when a range exists. The automatic listener snapshots `session.surface.replaceGeneration` and returns `{ kind: 'retry' }` whenever pruning or summarization increases it. This remains true when pruning lands before later summary work throws; cancellation still wins. A backend returning a result without replacement cannot authorize retry, while pruning-only progress can authorize a retry without a `CompactionResult`.
 

.agents/notes/implemented/architecture/2026-07-10-after-call-compaction-pressure-and-overflow-recovery.zh.md

@@ -32,7 +32,7 @@ Compact-basic 会在每个拟议请求之前包装 `agent/pre-step`。在续步
 
 `CompactService.compactIfNeeded(agent, trigger, signal)` 接收 `trigger: 'pressure' | 'context-overflow'`。接口不增加估算方法或 token 类型；`ctx.tokenMeter` 继续作为可复用的核算所有者。
 
-对于 `pressure`，compact-basic 先解析持久提供方/模型目标对应适配器所维护的容量与精确目标策略，再把得到的阈值与保留尾部预算应用到一次统一的 `ctx.tokenMeter.measure()` 结果。未达到压力阈值时直接返回，不执行剪枝。压力达到条件后，可选的 `ctx.toolResultPrune` 会改写当前表层中过大的工具结果，compact-basic 再通过同一个 meter 重新计量；若压力已降至安全水平则跳过模型调用，否则从已剪枝表层选择范围并生成摘要。范围定价、来源、被遮蔽 token 数与非缩小摘要拒绝也由同一个单例 meter 完成。通用默认值保持为阈值比例 `0.8`、保留历史比例 `0.16`、摘要提供方/模型 `''`、`maxTokens: 8192`、`compactionRetries: 1` 与 `auto: true`；可选 `modelPolicies` 项可以按精确提供方/模型组合覆盖这些值。
+对于 `pressure`，compact-basic 先解析持久提供方/模型目标对应适配器所维护的容量与精确目标策略，再把得到的阈值与保留尾部预算应用到一次统一的 `ctx.tokenMeter.measure()` 结果。未达到压力阈值时直接返回，不执行剪枝。压力达到条件后，可选的 `ctx.toolResultPrune` 会改写当前表层中过大的工具结果，compact-basic 再通过同一个 meter 重新计量；若压力已降至安全水平则跳过模型调用，否则从已剪枝表层选择范围并生成摘要。范围定价、引用的源事件计量、被遮蔽 token 数与非缩小摘要拒绝也由同一个单例 meter 完成。通用默认值保持为阈值比例 `0.8`、保留历史比例 `0.16`、摘要提供方/模型 `''`、`maxTokens: 8192`、`compactionRetries: 1` 与 `auto: true`；可选 `modelPolicies` 项可以按精确提供方/模型组合覆盖这些值。
 
 对于规范化溢出，compact-basic 不要求容量元数据，并绕过标量压力与普通保留 token 预算。它先执行剪枝，再在保留最新不可分割单元的同时选择最大的工具配对平衡头部范围；存在范围时，才在同一 signal 下尝试一次缩小摘要压缩。自动监听器先记录 `session.surface.replaceGeneration`，剪枝或摘要让 generation 增加时就返回 `{ kind: 'retry' }`。即使剪枝先落盘而后续摘要工作抛错，这条规则仍然成立；取消依然优先。后端若只返回结果但没有替换表层，不能授权重试；只有剪枝取得进展时，即使没有 `CompactionResult` 也可以授权重试。
 

.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.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 .agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md
-2026-07-14-provider-routed-llm-adapters.md: 9039334370ba5d71eb71879970c2e572a5b023df
-2026-07-14-provider-routed-llm-adapters.zh.md: eff33496e1a61472534009533270f8f943ccc1f8
+2026-07-14-provider-routed-llm-adapters.md: bb516a39b6c7303d125575bbd3d4138f9af65f7d
+2026-07-14-provider-routed-llm-adapters.zh.md: 2b6af0d509c28cb79dcd3342a007be8e54799ebc

.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md

@@ -10,7 +10,7 @@ English | [中文](2026-07-14-provider-routed-llm-adapters.zh.md)
 
 The conflation prevents a provider gateway from serving an open-ended model catalog. OpenRouter, for example, is one provider with many model ids, while a private OpenAI-compatible endpoint may add models without changing the Harness plugin tree. Every newly selected model currently needs to have been registered during plugin startup. The same model id can also exist at multiple providers, so model-only registration cannot state which provider the caller intended.
 
-`dsh-llm-pi-ai` exposed none of pi-ai's provider abstraction. It constructed an inline DeepSeek `openai-completions` model, applied DeepSeek-specific payload patches, and stamped every replayed assistant message as DeepSeek. pi-ai itself has a provider/model catalog, selects APIs such as `openai-responses`, `anthropic-messages`, and `google-generative-ai`, and preserves provider-specific response ids and reasoning/tool signatures for later turns. The Harness conversion dropped that provenance, so simply replacing the inline model with a catalog lookup would have made same-model replay and cross-provider handoff incomplete.
+`dsh-llm-pi-ai` exposed none of pi-ai's provider abstraction. It constructed an inline DeepSeek `openai-completions` model, applied DeepSeek-specific payload patches, and stamped every replayed assistant message as DeepSeek. pi-ai itself has a provider/model catalog, selects APIs such as `openai-responses`, `anthropic-messages`, and `google-generative-ai`, and preserves provider-specific response ids and reasoning/tool signatures for later turns. The Harness conversion dropped the provider/model route and provider response fields, so simply replacing the inline model with a catalog lookup would have made same-model replay and cross-provider handoff incomplete.
 
 The adapter configuration also assumes one DeepSeek API key and endpoint. A generic backend needs independent credentials and endpoint overrides per provider while leaving AWS, Google ADC, OAuth, and other ambient authentication mechanisms to pi-ai.
 
@@ -36,15 +36,15 @@ The adapter calls pi-ai's `streamSimple()` so each catalog model chooses its reg
 
 pi-ai's common stream options do not expose stop sequences. `dsh-llm-pi-ai` rejects a defined Harness `stop` option with `UNSUPPORTED_OPTION` rather than silently ignoring it or growing a second provider-specific payload implementation. `dsh-llm-deepseek` continues to support `stop` through its native request serializer.
 
-### Durable assistant provenance and replay state
+### Recorded assistant route and replay state
 
-Assistant messages carry provider-neutral provenance containing the request's `provider` and `model`, plus an optional JSON-serializable adapter replay state. A successful `assistant/message` session event records this provenance and `deriveMessages()` returns it with the assistant message. User, system, context, and tool-result messages carry no assistant provenance. The provider/model fields are authoritative loop data; an adapter owns only its opaque replay-state payload.
+Assistant messages carry the request's `provider` and `model`, plus an optional JSON-serializable adapter replay state. A successful `assistant/message` session event records those fields and `deriveMessages()` returns them with the assistant message. User, system, context, and tool-result messages carry no assistant route fields. The provider/model fields are authoritative loop data; an adapter owns only its opaque replay-state payload.
 
-A terminal successful `finish` chunk may carry replay state, and `BlockAssembler` retains it alongside usage and finish reason. The loop attaches that state to the assembled assistant provenance without exposing a response-rewrite hook. Error and aborted responses do not produce a normal assistant message and therefore do not enter future model history.
+A terminal successful `finish` chunk may carry replay state, and `BlockAssembler` retains it alongside usage and finish reason. The loop attaches that state to the assembled assistant message's model source without exposing a response-rewrite hook. Error and aborted responses do not produce a normal assistant message and therefore do not enter future model history.
 
-The pi-ai replay state is a versioned, minimal projection of its successful `AssistantMessage`: source API/provider/model, response id/model, stop reason, and index-aligned text, thinking, and tool-call signatures. It does not duplicate text or tool arguments already carried by Harness content blocks, and it omits diagnostics, timestamps, usage, and errors. On a later request, `LlmService` gives replay state to the target adapter only when the historical provider and target provider are currently owned by the same adapter instance. That adapter combines the logged Harness content with replay state when it can restore the historical response, and owns any required cross-model or cross-provider conversion. An adapter receiving replay state with an unknown version or mismatched block shape fails explicitly; a different adapter receives only provider-neutral content and provenance.
+The pi-ai replay state is a versioned, minimal projection of its successful `AssistantMessage`: source API/provider/model, response id/model, stop reason, and index-aligned text, thinking, and tool-call signatures. It does not duplicate text or tool arguments already carried by Harness content blocks, and it omits diagnostics, timestamps, usage, and errors. On a later request, `LlmService` gives replay state to the target adapter only when the historical provider and target provider are currently owned by the same adapter instance. That adapter combines the logged Harness content with replay state when it can restore the historical response, and owns any required cross-model or cross-provider conversion. An adapter receiving replay state with an unknown version or mismatched block shape fails explicitly; a different adapter receives only provider-neutral content plus provider/model fields.
 
-This state is model-visible replay input and therefore follows the existing [reconstructable-request rule](2026-07-05-reconstructable-requests.md): it is present in both the terminal `finish` chunk and the assembled `assistant/message` provenance that drives derivation. Resume and fork preserve it verbatim. Compaction that shadows the assistant message also removes its replay state from the active surface; the summary is ordinary provider-neutral content.
+This state is model-visible replay input and therefore follows the existing [reconstructable-request rule](2026-07-05-reconstructable-requests.md): it is present in both the terminal `finish` chunk and the assembled `assistant/message` model source that drives derivation. Resume and fork preserve it verbatim. Compaction that shadows the assistant message also removes its replay state from the active surface; the summary is ordinary provider-neutral content.
 
 ### Propagate the target through every request producer
 
@@ -54,7 +54,7 @@ Compaction configuration gains `summarizationProvider` beside `summarizationMode
 
 The JSON-RPC runtime receives provider and model explicitly. Its convenience fallback mounts `dsh-llm-deepseek` only for provider `deepseek` when that provider has no registered owner; other missing providers fail without guessing an adapter.
 
-The on-disk session format remains the pre-release pinned version `0`, with no compatibility promise. Seed/load validation rejects request headers lacking provider and assistant messages lacking required provenance instead of accepting an old shape that can no longer reconstruct the request.
+The on-disk session format remains the pre-release pinned version `0`, with no compatibility promise. Seed/load validation rejects request headers and assistant messages that omit required provider/model fields instead of accepting an old shape that can no longer reconstruct the request.
 
 ## Alternatives considered
 
@@ -78,7 +78,7 @@ The on-disk session format remains the pre-release pinned version `0`, with no c
 - pi-ai credentials, transport knobs, SDK timeouts, and the five-minute-default `streamIdleTimeoutMs` watchdog are scoped per provider profile. Hidden provider retries are disabled; bounded retries belong to the separately composed agent recovery policy.
 - `dsh-llm-pi-ai` rejects stop sequences because pi-ai's common stream API cannot express them; the native DeepSeek adapter retains its stop support.
 - Replay state is portable only within the adapter instance that owns both the historical and target providers. Cross-provider and cross-model restoration is an adapter responsibility, and another adapter receives provider-neutral history without the opaque state.
-- Current pre-release session JSONL requires provider/model request headers and assistant provenance. Older shapes remain version `0` but are rejected rather than migrated.
+- Current pre-release session JSONL requires provider/model on request headers and assistant messages. Older shapes remain version `0` but are rejected rather than migrated.
 
 ## Testing
 

.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md

@@ -10,7 +10,7 @@ Status: implemented
 
 这种混淆使提供方网关无法提供开放的模型目录。例如，OpenRouter 是一个包含大量模型 ID 的提供方，私有 OpenAI 兼容端点也可能在不修改 Harness 插件树的情况下增加模型。目前，每个新选择的模型都必须在插件启动期间完成注册。同一个模型 ID 还可能存在于多个提供方中，因此仅按模型注册无法表达调用方预期使用的提供方。
 
-`dsh-llm-pi-ai` 没有暴露 pi-ai 的提供方抽象。它以内联方式构造 DeepSeek `openai-completions` 模型，应用 DeepSeek 专用的 payload 补丁，并将每条回放的助手消息标记为 DeepSeek。pi-ai 自身提供提供方/模型目录，能够选择 `openai-responses`、`anthropic-messages`、`google-generative-ai` 等 API，并保留提供方专用的响应 ID，以及后续轮次所需的推理和工具签名。Harness 转换丢弃了这些来源信息，因此仅将内联模型替换为目录查询，会导致同模型回放与跨提供方移交不完整。
+`dsh-llm-pi-ai` 没有暴露 pi-ai 的提供方抽象。它以内联方式构造 DeepSeek `openai-completions` 模型，应用 DeepSeek 专用的 payload 补丁，并将每条回放的助手消息标记为 DeepSeek。pi-ai 自身提供提供方/模型目录，能够选择 `openai-responses`、`anthropic-messages`、`google-generative-ai` 等 API，并保留提供方专用的响应 ID，以及后续轮次所需的推理和工具签名。Harness 转换丢弃了提供方／模型路由和提供方响应字段，因此仅将内联模型替换为目录查询，会导致同模型回放与跨提供方移交不完整。
 
 适配器配置同样假定只存在一个 DeepSeek API 密钥和端点。通用后端需要为各提供方分别配置凭据和端点覆盖，同时继续由 pi-ai 处理 AWS、Google ADC、OAuth 等环境认证机制。
 
@@ -36,15 +36,15 @@ Status: implemented
 
 pi-ai 的通用流选项不支持停止序列。若 Harness `stop` 选项已定义，`dsh-llm-pi-ai` 会以 `UNSUPPORTED_OPTION` 拒绝请求，不会静默忽略，也不会增加第二套提供方专用 payload 实现。`dsh-llm-deepseek` 继续通过原生请求序列化器支持 `stop`。
 
-### 持久化助手来源信息与回放状态
+### 已记录的助手路由与回放状态
 
-助手消息携带提供方无关的来源信息，其中包含请求的 `provider` 和 `model`，以及可选的 JSON 可序列化适配器回放状态。成功的 `assistant/message` 会话事件记录这些来源信息，`deriveMessages()` 返回助手消息时也会包含这些信息。用户、system、context 与工具结果消息不携带助手来源信息。provider/model 字段是 agent loop 的权威数据；适配器仅拥有其不透明回放状态 payload。
+助手消息携带请求的 `provider` 和 `model`，以及可选的 JSON 可序列化适配器回放状态。成功的 `assistant/message` 会话事件记录这些字段，`deriveMessages()` 返回助手消息时也会包含它们。用户、system、context 与工具结果消息不携带助手路由字段。provider/model 字段是 agent loop 的权威数据；适配器仅拥有其不透明回放状态 payload。
 
-成功的终止 `finish` 分片可以携带回放状态，`BlockAssembler` 会将其与 token 用量和结束原因一起保留。只有当 `agent/step-result` 处理后的内容与提供方组装输出在结构上相等时，agent loop 才会把回放状态附加到助手来源信息。监听器重写内容后，provider/model 来源信息仍会保留，但已经陈旧的回放状态会被移除。错误或中止响应不会生成正常助手消息，因此不会进入后续模型历史。
+成功的终止 `finish` 分片可以携带回放状态，`BlockAssembler` 会将其与 token 用量和结束原因一起保留。agent loop 会把该状态附加到已组装助手消息的模型来源中，但不公开响应改写钩子。错误或中止响应不会生成正常助手消息，因此不会进入后续模型历史。
 
-pi-ai 回放状态是其成功 `AssistantMessage` 的带版本最小投影，包含源 API/provider/model、响应 ID/model、停止原因，以及按索引对齐的文本、thinking 和工具调用签名。它不会重复 Harness 内容块中已有的文本或工具参数，也不包含诊断信息、时间戳、用量或错误。后续请求中，只有历史提供方和目标提供方当前归同一个适配器实例所有时，`LlmService` 才会把回放状态交给目标适配器。适配器在能够恢复历史响应时，将 Harness 记录的内容与回放状态组合，并负责所需的跨模型或跨提供方转换。适配器收到未知版本或块形状不匹配的回放状态时会显式失败；其他适配器只能收到提供方无关的内容与来源信息。
+pi-ai 回放状态是其成功 `AssistantMessage` 的带版本最小投影，包含源 API/provider/model、响应 ID/model、停止原因，以及按索引对齐的文本、thinking 和工具调用签名。它不会重复 Harness 内容块中已有的文本或工具参数，也不包含诊断信息、时间戳、用量或错误。后续请求中，只有历史提供方和目标提供方当前归同一个适配器实例所有时，`LlmService` 才会把回放状态交给目标适配器。适配器在能够恢复历史响应时，将 Harness 记录的内容与回放状态组合，并负责所需的跨模型或跨提供方转换。适配器收到未知版本或块形状不匹配的回放状态时会显式失败；其他适配器只能收到提供方无关的内容以及 provider/model 字段。
 
-该状态属于模型可见的回放输入，因此遵循现有的[请求可重建规则](2026-07-05-reconstructable-requests.md)：它同时存在于终止 `finish` 分片和驱动派生的已组装 `assistant/message` 来源信息中。恢复和 fork 会原样保留该状态。压缩（compaction）遮蔽助手消息时，也会从活动 surface 中移除其回放状态；摘要属于普通的提供方无关内容。
+该状态属于模型可见的回放输入，因此遵循现有的[请求可重建规则](2026-07-05-reconstructable-requests.md)：它同时存在于终止 `finish` 分片和驱动派生的已组装 `assistant/message` 模型来源中。恢复和 fork 会原样保留该状态。压缩（compaction）遮蔽助手消息时，也会从活动 surface 中移除其回放状态；摘要属于普通的提供方无关内容。
 
 ### 在所有请求生产方中传播目标
 
@@ -54,7 +54,7 @@ pi-ai 回放状态是其成功 `AssistantMessage` 的带版本最小投影，包
 
 JSON-RPC 运行时显式接收 provider 与 model。仅当 `deepseek` 提供方没有注册所有者时，其便利回退才会挂载 `dsh-llm-deepseek`；其他缺失的提供方会直接失败，不会猜测适配器。
 
-磁盘会话格式仍使用预发布阶段固定的版本 `0`，且不承诺兼容性。seed/load 验证会拒绝缺少 provider 的请求头，以及缺少必需来源信息的助手消息，不会接受已无法重建请求的旧格式。
+磁盘会话格式仍使用预发布阶段固定的版本 `0`，且不承诺兼容性。seed/load 验证会拒绝省略必需 provider/model 字段的请求头和助手消息，不会接受已无法重建请求的旧格式。
 
 ## 考虑过的替代方案
 
@@ -78,7 +78,7 @@ JSON-RPC 运行时显式接收 provider 与 model。仅当 `deepseek` 提供方
 - pi-ai 凭据、传输选项、SDK 超时，以及默认五分钟的 `streamIdleTimeoutMs` 空闲超时机制均按提供方配置隔离。系统禁用隐藏的提供方重试；有界重试由单独组合的 agent 恢复策略负责。
 - pi-ai 的通用流 API 无法表达停止序列，因此 `dsh-llm-pi-ai` 会拒绝停止序列；原生 DeepSeek 适配器仍支持停止序列。
 - 仅当历史提供方与目标提供方归同一个适配器实例所有时，回放状态才可移植。适配器负责跨提供方和跨模型恢复；其他适配器只接收不含不透明状态的提供方无关历史。
-- 当前预发布会话 JSONL 要求请求头包含 provider/model，助手消息包含来源信息。旧格式仍使用版本 `0`，但会被拒绝，不执行迁移。
+- 当前预发布会话 JSONL 要求请求头和助手消息都包含 provider/model。旧格式仍使用版本 `0`，但会被拒绝，不执行迁移。
 
 ## 测试
 

.agents/notes/implemented/architecture/2026-07-15-replay-token-meter-service.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 .agents/notes/implemented/architecture/2026-07-15-replay-token-meter-service.md
-2026-07-15-replay-token-meter-service.md: c0f4b467ad0013dd4ac0a0301281b011ea8c261c
-2026-07-15-replay-token-meter-service.zh.md: c1d81dc0e76ee687ced5c23d26197627551e1ced
+2026-07-15-replay-token-meter-service.md: 4013cc92f67597a9b87edfc97d14c7f47f0523ac
+2026-07-15-replay-token-meter-service.zh.md: 4bee27db2a7d021b25dfdf1feefa148cfb016100

.agents/notes/implemented/architecture/2026-07-15-replay-token-meter-service.md

@@ -8,7 +8,7 @@ English | [中文](2026-07-15-replay-token-meter-service.zh.md)
 
 Context pressure is useful outside compaction. A compaction backend, an overflow guard, or a future request-policy plugin can all need the same answer: how many tokens does the durable request consume? Keeping that fold inside `dsh-compact-basic` duplicates replay logic, makes measurement unavailable without compaction, and encourages callers to reuse stale accounting.
 
-Provider usage is not a complete answer. It describes one successful call under one exact request envelope, while the current surface can grow, shrink, or be replaced afterward. Sessions also switch providers and models, old logs can lack chunk provenance, and usage fields separate input, cache-read, cache-write, output, and reasoning counts. A useful service therefore combines the latest exact anchor with conservative heuristic repricing and exposes the log revision consumed by each result.
+Provider usage is not a complete answer. It describes one successful call under one exact request envelope, while the current surface can grow, shrink, or be replaced afterward. Sessions also switch providers and models, old logs can omit the chunk seqs behind an assistant message, and usage fields separate input, cache-read, cache-write, output, and reasoning counts. A useful service therefore combines the latest exact anchor with conservative heuristic repricing and exposes the log revision consumed by each result.
 
 ## Decision
 
@@ -20,17 +20,17 @@ The service has no configuration. Estimation uses a fixed four-characters-per-to
 
 ### Per-session replay folds
 
-Each session owns one isolated incremental fold. Active folds advance from `session/event`; every read catches up through the durable tail, so listener ordering, seeded sessions, and service reload do not change the answer. The fold tracks canonical full request-header snapshots, step boundaries, surface appends and replacements, assistant usage, and assistant-chunk provenance. A malformed next event fails transactionally and remains unread rather than partially mutating state.
+Each session owns one isolated incremental fold. Active folds advance from `session/event`; every read catches up through the durable tail, so listener ordering, seeded sessions, and service reload do not change the answer. The fold tracks canonical full request-header snapshots, step boundaries, surface appends and replacements, assistant usage, and the chunk seqs cited by each assistant message. A malformed next event fails transactionally and remains unread rather than partially mutating state.
 
 `measure(session, requestHeader?)` synchronizes the fold once and returns scalar pressure together with positional per-node prices. `totalTokens` remains request-and-response pressure; `surfaceTokens` is the surface-only heuristic total and equals the sum of `nodes[].tokens`. A `requestHeader` override changes pressure pricing only, while the surface fields always describe the current session. `estimateMessage(message)` applies the fixed heuristic without session state. Each result is one detached, deeply immutable snapshot carrying one `logRevision`. Every measurement clones the current nodes and is therefore O(surface).
 
 Provider usage is reused only when the measured canonical request envelope equals the latest successful-call anchor. Any provider, model, system, prefix, tool, or call-config change causes complete heuristic repricing. Surface changes remain a signed delta from a matching anchor, including negative values after a shrinking replacement. A later successful request replaces the earlier anchor, including across provider or model switches.
 
-Usage sums the disjoint input, cache-read, cache-write, and output buckets. Reasoning is not added a second time. Every successful model call records an `assistant/message`, including content-less and max-token calls, with its exact earlier chunk seqs. An explicit empty provenance list means a known empty provider stream; absent legacy provenance conservatively treats the durable assistant output as provider output.
+Usage sums the disjoint input, cache-read, cache-write, and output buckets. Reasoning is not added a second time. Every successful model call records an `assistant/message`, including content-less and max-token calls, with its exact earlier chunk seqs. An explicit empty `sourceEventSeqs` list means a known empty provider stream; an absent legacy list conservatively treats the durable assistant output as provider output.
 
 ### Compact-basic consumes, but does not own, measurement
 
-`dsh-compact-basic` requires `ctx.tokenMeter`; `CompactService` gains no token methods or types. Configuration, the region transaction, and summarization stay in separate modules; the service registers automatic listeners itself, while `summarize()` remains its sole subclass hook. The singleton meter consistently prices pressure, retention, shadowed content, provenance, and non-shrinking-summary rejection.
+`dsh-compact-basic` requires `ctx.tokenMeter`; `CompactService` gains no token methods or types. Configuration, the region transaction, and summarization stay in separate modules; the service registers automatic listeners itself, while `summarize()` remains its sole subclass hook. The singleton meter consistently prices pressure, retention, shadowed content, cited source events, and non-shrinking-summary rejection.
 
 Automatic compaction uses one unified measurement for each threshold-and-retention decision. The region transaction measures after appending its durable `compact/start` lock and again after asynchronous summarization, then compares the detached surface-node vectors. An intervening surface mutation prevents replacement; `logRevision` may advance for unrelated log-only facts without invalidating an unchanged selected span.
 

.agents/notes/implemented/architecture/2026-07-15-replay-token-meter-service.zh.md

@@ -8,7 +8,7 @@ Status: implemented
 
 上下文压力并不只对压缩（compaction）有用。压缩后端、溢出保护或未来的请求策略插件都可能需要回答同一个问题：持久请求消耗了多少 token？如果把该折叠逻辑留在 `dsh-compact-basic` 内部，就会重复实现回放逻辑，使未加载压缩的调用方无法使用计量，并诱使调用方复用陈旧的核算结果。
 
-提供方 usage 也不是完整答案。它只描述某个精确请求信封下的一次成功调用，而当前表层之后还可能增长、缩小或被替换。会话也可能切换提供方与模型，旧日志可能缺少分片来源，usage 字段还会分别报告输入、缓存读取、缓存写入、输出与推理计数。因此，可用的服务必须把最新精确锚点与保守的启发式重新定价结合起来，并公开每个结果已经消费的日志修订号。
+提供方 usage 也不是完整答案。它只描述某个精确请求信封下的一次成功调用，而当前表层之后还可能增长、缩小或被替换。会话也可能切换提供方与模型，旧日志可能缺少构成 assistant 消息的分片 seq，usage 字段还会分别报告输入、缓存读取、缓存写入、输出与推理计数。因此，可用的服务必须把最新精确锚点与保守的启发式重新定价结合起来，并公开每个结果已经消费的日志修订号。
 
 ## 决策
 
@@ -20,17 +20,17 @@ Status: implemented
 
 ### 逐会话回放折叠
 
-每个会话都有一个隔离的增量折叠。活跃折叠通过 `session/event` 前进；每次读取都会追到持久日志尾部，因此监听器顺序、种子会话与服务重载不会改变答案。折叠跟踪规范的完整请求头快照、步骤边界、表层追加与替换、assistant usage，以及 assistant 分片来源。下一个畸形事件会以事务方式失败并保持未读，不会让状态只修改一半。
+每个会话都有一个隔离的增量折叠。活跃折叠通过 `session/event` 前进；每次读取都会追到持久日志尾部，因此监听器顺序、种子会话与服务重载不会改变答案。折叠跟踪规范的完整请求头快照、步骤边界、表层追加与替换、assistant usage，以及每条 assistant 消息引用的分片 seq。下一个畸形事件会以事务方式失败并保持未读，不会让状态只修改一半。
 
 `measure(session, requestHeader?)` 只同步一次折叠，并在返回标量压力的同时给出逐位置节点价格。`totalTokens` 仍表示请求与响应压力；`surfaceTokens` 是仅针对表层的启发式总量，并等于 `nodes[].tokens` 之和。`requestHeader` 覆盖只改变压力定价，表层字段始终描述当前会话。`estimateMessage(message)` 不依赖会话状态，直接应用固定启发式规则。每个结果都是一个分离且深度不可变的快照，只携带一个 `logRevision`。每次计量都会复制当前节点，因此成本为 O(surface)。
 
 只有当待计量的规范请求信封等于最近一次成功调用的锚点时，服务才复用提供方 usage。提供方、模型、系统提示词、前缀、工具或调用配置任一变化都会触发完整的启发式重新定价。表层变化相对匹配锚点保留有符号增量，包括缩小替换后的负值。后续成功请求会替换先前锚点，提供方或模型切换时也一样。
 
-Usage 会对互不重叠的输入、缓存读取、缓存写入与输出 bucket 求和，不会再次加入推理计数。每次成功模型调用都会记录 `assistant/message`，包括无内容调用与达到 token 上限的调用，并带上精确的更早分片 seq。显式空来源列表表示已知为空的提供方流；旧日志中缺失的来源则保守地把持久 assistant 输出视为提供方输出。
+Usage 会对互不重叠的输入、缓存读取、缓存写入与输出 bucket 求和，不会再次加入推理计数。每次成功模型调用都会记录 `assistant/message`，包括无内容调用与达到 token 上限的调用，并带上精确的更早分片 seq。显式的空 `sourceEventSeqs` 列表表示已知为空的提供方流；旧日志中缺失的列表则保守地把持久 assistant 输出视为提供方输出。
 
 ### compact-basic 消费计量，但不拥有计量
 
-`dsh-compact-basic` 要求 `ctx.tokenMeter`；`CompactService` 不增加 token 方法或类型。配置、区域事务与摘要分别保留在独立模块中，服务自身注册自动监听器，而 `summarize()` 仍是唯一的子类钩子。单例计量器一致用于压力、保留、被遮蔽内容、来源以及非缩小摘要拒绝的定价。
+`dsh-compact-basic` 要求 `ctx.tokenMeter`；`CompactService` 不增加 token 方法或类型。配置、区域事务与摘要分别保留在独立模块中，服务自身注册自动监听器，而 `summarize()` 仍是唯一的子类钩子。单例计量器一致用于压力、保留、被遮蔽内容、引用的源事件以及非缩小摘要拒绝的定价。
 
 自动压缩的每次阈值与保留联合决策只使用一次统一计量。区域事务会在追加持久 `compact/start` 锁后执行计量，在异步摘要完成后再次计量，随后比较分离的表层节点向量。期间发生的表层变更会阻止替换；`logRevision` 可以因无关的纯日志事实而推进，而不会使未变的选定范围失效。
 

.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.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 .agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.md
-2026-07-19-package-owned-invariant-service.md: e32efe9f6b3ce6b782c61db56d928e87c160dc9a
-2026-07-19-package-owned-invariant-service.zh.md: abb77cfc2dc4c15510baf0b205cefdbf3c890041
+2026-07-19-package-owned-invariant-service.md: 6ab53f1a9bfa043f17875fa179a99b832b3fcf1a
+2026-07-19-package-owned-invariant-service.zh.md: 378b730c479eed05a2ad11e6a6e084929ec09653

.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.md

@@ -102,4 +102,4 @@ Every Vitest configuration loads a test host that mounts an explicitly enabled s
 - One selected executable contribution adds one child fiber and its listener/state cost; a selected empty contribution has no listener or trace-state cost, while filtered registrations retain only name ownership.
 - Regex sources are deployment configuration and remain fixed until the service reloads.
 - Ordinary Vitest roots install the owning test package's selected companion; one exhaustive topology pays the full child-fiber cost once for repository-wide registration coverage.
-- Session storage validation, snapshotting, freezing, provenance, and surface acceptance remain always on and are not affected by invariant selection.
+- Session storage validation, snapshotting, freezing, cited source-event validation, and surface acceptance remain always on and are not affected by invariant selection.

.agents/notes/implemented/architecture/2026-07-19-package-owned-invariant-service.zh.md

@@ -102,4 +102,4 @@ Workspace 约束识别独立的不变式 bundle；包 exports、项目引用、
 - 每个选中的可执行贡献增加一个子 fiber 及其 listener/状态成本；选中的空贡献不增加 listener 或 trace 状态成本，被过滤注册则只保留包名占用。
 - 正则表达式源属于部署配置，在服务重载前保持固定。
 - 普通 Vitest 根上下文会安装当前测试包中被选中的伴随插件；一个完整拓扑只支付一次全部子 fiber 成本，用于覆盖整个仓库的注册。
-- 会话存储验证、快照、冻结、provenance 与 surface 接受规则始终启用，不受不变式选择影响。
+- 会话存储验证、快照、冻结、引用的源事件验证与 surface 接受规则始终启用，不受不变式选择影响。

.agents/notes/implemented/architecture/2026-07-20-canonical-tool-output-contract.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 .agents/notes/implemented/architecture/2026-07-20-canonical-tool-output-contract.md
-2026-07-20-canonical-tool-output-contract.md: 2429e1c141ad8c8ee932c6d654a6ba74bc4f7618
-2026-07-20-canonical-tool-output-contract.zh.md: 534aa2ed7ad65719b280f64f73a412bf5b17a1de
+2026-07-20-canonical-tool-output-contract.md: fc4a8d52f2ad6ff532fafdd7a987a095c8e3bcef
+2026-07-20-canonical-tool-output-contract.zh.md: a42c5d9a2b581012af823343991ac5789a787658

.agents/notes/implemented/architecture/2026-07-20-canonical-tool-output-contract.md

@@ -24,7 +24,7 @@ output: {
 
 `defineTool` infers the body return and both projectors from the unified `ValueSchemaSpec`. Raw and dynamic definitions provide the compiled `JsonSchemaNode` form. Registration rejects a missing declaration or unsupported raw schema; there is no content-return compatibility path.
 
-For each successful dispatch the registry snapshots the returned value as lossless `JsonValue`, validates it against `output.schema`, deep-freezes it, then invokes the pure renderer and, for a direct surface call, the optional metadata projector. Renderer, projector, schema, or lossless-JSON failures are contained as ordinary `ToolOutputError` results. An around `tools/execute` wrapper receives and returns the canonical success/failure union; a wrapper-authored success is normalized again through the resolved tool's output declaration instead of trusting independently authored content. Canonical-result provenance is scoped to the immutable dispatch token, so returning a cached result from another call or tool triggers normalization under the active declaration rather than bypassing it.
+For each successful dispatch the registry snapshots the returned value as lossless `JsonValue`, validates it against `output.schema`, deep-freezes it, then invokes the pure renderer and, for a direct surface call, the optional metadata projector. Renderer, projector, schema, or lossless-JSON failures are contained as ordinary `ToolOutputError` results. An around `tools/execute` wrapper receives and returns the canonical success/failure union; a wrapper-authored success is normalized again through the resolved tool's output declaration instead of trusting independently authored content. Each canonical result is tied to the immutable dispatch token that created it, so returning a cached result from another call or tool triggers normalization under the active declaration rather than bypassing it.
 
 ```ts ignore-check
 type ToolExecutionResult =

.agents/notes/implemented/architecture/2026-07-20-canonical-tool-output-contract.zh.md

@@ -24,7 +24,7 @@ output: {
 
 `defineTool` 从统一的 `ValueSchemaSpec` 推导工具主体返回值和两个投影器的类型。原始定义和动态定义则提供编译后的 `JsonSchemaNode` 形式。注册时会拒绝缺失输出声明或采用不受支持的原始 schema 的定义，不提供兼容旧式内容返回值的路径。
 
-每次成功分发时，注册表会将返回值快照为无损 `JsonValue`，依据 `output.schema` 校验并深度冻结，然后调用纯渲染器；对于直接的外层调用，还会调用可选的元数据投影器。渲染器、投影器、schema 或无损 JSON 处理失败都会被收敛为普通 `ToolOutputError` 结果。围绕 `tools/execute` 的包装层接收并返回规范的成功／失败联合；包装层自行产生的成功结果会再次通过已解析工具的输出声明完成归一化，而不会信任其独立编写的内容。每个规范结果的来源归属仅限于一个不可变的分发 token；因此，如果包装层返回来自其他调用或工具的缓存结果，系统会依据当前生效的输出声明重新执行归一化，而不会绕过这一步。
+每次成功分发时，注册表会将返回值快照为无损 `JsonValue`，依据 `output.schema` 校验并深度冻结，然后调用纯渲染器；对于直接的外层调用，还会调用可选的元数据投影器。渲染器、投影器、schema 或无损 JSON 处理失败都会被收敛为普通 `ToolOutputError` 结果。围绕 `tools/execute` 的包装层接收并返回规范的成功／失败联合；包装层自行产生的成功结果会再次通过已解析工具的输出声明完成归一化，而不会信任其独立编写的内容。每个规范结果都与创建它的不可变分发 token 绑定；因此，如果包装层返回来自其他调用或工具的缓存结果，系统会依据当前生效的输出声明重新执行归一化，而不会绕过这一步。
 
 ```ts ignore-check
 type ToolExecutionResult =

.agents/notes/implemented/architecture/2026-07-22-unified-send-and-coalesced-user-messages.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 .agents/notes/implemented/architecture/2026-07-22-unified-send-and-coalesced-user-messages.md
-2026-07-22-unified-send-and-coalesced-user-messages.md: f32d6ca65d5236e1fabdd177cdf54e36929c853f
-2026-07-22-unified-send-and-coalesced-user-messages.zh.md: 6148cf4e0dda9032fca0ced5b466787877b89fa0
+2026-07-22-unified-send-and-coalesced-user-messages.md: bc195e77c71b554d60c03c91134bbcef95678fff
+2026-07-22-unified-send-and-coalesced-user-messages.zh.md: 668ca01bc6b6854642fee2afe9f7be7354b25c5c

.agents/notes/implemented/architecture/2026-07-22-unified-send-and-coalesced-user-messages.md

@@ -14,9 +14,9 @@ Separately, `context/message` and `user/message` had converged: the surface proj
 
 **One primitive, three preset aliases.** The `Agent` interface's `send(message, target, wakeup)` covers the (`target` × `wakeup`) matrix. Its complete `UserMessage` owns identity, role, model-facing `content`, and producer `source`; the remaining arguments own only routing policy. `followup` (`next-turn`/wakeup), `steer` (`next-step`/wakeup), and `inject` (`next-step`/no-wakeup) each accept that one message and fix the policy. `wakeup` reserves a driver when the agent is idle; an already active driver receives no second reservation and can claim the input only if it reaches a later pre-step boundary. `next-turn`/no-wakeup (queue without waking) is representable with no alias and no current caller.
 
-**inject is a non-waking next-step delivery.** It always appends the complete message to the next-step inbox and records that insertion in a durable `agent/inbox/spliced` event. The driver claims it at a later pre-step and records it as model-visible `user/message` only when the final decision returns it in the entering batch; idle injection remains pending until another delivery wakes the driver. Its required `UserMessage.source` preserves the caller's explicit provenance.
+**inject is a non-waking next-step delivery.** It always appends the complete message to the next-step inbox and records that insertion in a durable `agent/inbox/spliced` event. The driver claims it at a later pre-step and records it as model-visible `user/message` only when the final decision returns it in the entering batch; idle injection remains pending until another delivery wakes the driver. Its required `UserMessage.source` preserves the source fields supplied by the caller.
 
-**context/message is gone.** Injected context uses one `UserMessage` value in the inbox and becomes a `user/message` event if admitted; context producers supply the appropriate non-user `source` explicitly, and typed source variants carry any domain-specific durable provenance. The surface, derivation, and `SurfaceEventType` drop `context/message`; consumers that need "is this a human prompt?" read `source.kind === 'user'` instead of the event type.
+**context/message is gone.** Injected context uses one `UserMessage` value in the inbox and becomes a `user/message` event if admitted; context producers supply the appropriate non-user `source` explicitly, and typed source variants carry any durable producer-specific fields. The surface, derivation, and `SurfaceEventType` drop `context/message`; consumers that need "is this a human prompt?" read `source.kind === 'user'` instead of the event type.
 
 **Goal continuation attribution uses positive rounds.** Goal lifecycle state commits through the domain-owned `goal/change` event defined by the later [goal-owned durable event decision](2026-07-31-goal-owned-durable-events.md). A positive round advances only from an admitted continuation `user/message`; goal persistence does not use injection or inbox state.
 

.agents/notes/implemented/architecture/2026-07-22-unified-send-and-coalesced-user-messages.zh.md

@@ -14,9 +14,9 @@ agent 的对外驱动接口逐渐长出三个近乎平行的动词——`send`
 
 **一个原语，三个预设别名。** `Agent` 接口的 `send(message, target, wakeup)` 覆盖 (`target` × `wakeup`) 矩阵。完整的 `UserMessage` 持有标识、角色、模型可见 `content` 与生产方 `source`；其余参数只持有路由策略。`followup`（`next-turn`/wakeup）、`steer`（`next-step`/wakeup）和 `inject`（`next-step`/no-wakeup）都接收这一条消息并固定策略。`wakeup` 会在 agent 空闲时保留一个驱动器；已经活跃的驱动器不会获得第二次保留，只有在抵达后续 pre-step 边界时才能领取该输入。`next-turn`/no-wakeup（入队但不唤醒）可以表达，只是没有别名，也没有当前调用方。
 
-**inject 是不会唤醒的 next-step 投递。** 它始终把完整消息追加到 next-step inbox，并在持久 `agent/inbox/spliced` 事件中记录该插入。驱动器会在后续 pre-step 领取它，并且只有最终决策把它放入进入步骤的批次时，才会将其记录为模型可见的 `user/message`；空闲注入会保持待处理，直到其他投递唤醒驱动器。必填的 `UserMessage.source` 会保留调用方显式提供的来源信息。
+**inject 是不会唤醒的 next-step 投递。** 它始终把完整消息追加到 next-step inbox，并在持久 `agent/inbox/spliced` 事件中记录该插入。驱动器会在后续 pre-step 领取它，并且只有最终决策把它放入进入步骤的批次时，才会将其记录为模型可见的 `user/message`；空闲注入会保持待处理，直到其他投递唤醒驱动器。必填的 `UserMessage.source` 会保留调用方提供的源字段。
 
-**context/message 已移除。** 注入的上下文在 inbox 中使用同一个 `UserMessage` 值，并在获准时成为 `user/message` 事件；上下文生产方显式提供合适的非 `user` 类别 `source`，类型化 source 变体携带所有特定于领域的持久来源信息。对外接口、派生逻辑和 `SurfaceEventType` 都不再包含 `context/message`；需要判断“这是不是一条人类提示词？”的消费方改为读取 `source.kind === 'user'`，而不是事件类型。
+**context/message 已移除。** 注入的上下文在 inbox 中使用同一个 `UserMessage` 值，并在获准时成为 `user/message` 事件；上下文生产方显式提供合适的非 `user` 类别 `source`，类型化 source 变体携带持久化的生产方专用字段。对外接口、派生逻辑和 `SurfaceEventType` 都不再包含 `context/message`；需要判断“这是不是一条人类提示词？”的消费方改为读取 `source.kind === 'user'`，而不是事件类型。
 
 **Goal 继续执行归属使用正数 Round。** Goal 生命周期状态通过后续的[Goal 自有持久事件决策](2026-07-31-goal-owned-durable-events.md)所定义的领域自有 `goal/change` 事件提交。正数 Round 只从已准入的继续执行 `user/message` 推进；goal 持久化不使用注入或 inbox 状态。
 

.agents/notes/implemented/architecture/2026-07-24-separate-context-injection-from-turn-execution.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 .agents/notes/implemented/architecture/2026-07-24-separate-context-injection-from-turn-execution.md
-2026-07-24-separate-context-injection-from-turn-execution.md: bb28d96cf1494d94ad7a7d4a4714e536c7072d2f
-2026-07-24-separate-context-injection-from-turn-execution.zh.md: 0edf3f4e8a062fe82462e263da383e27d79034c5
+2026-07-24-separate-context-injection-from-turn-execution.md: 719c4562b6687e42cc70721658152992735ec662
+2026-07-24-separate-context-injection-from-turn-execution.zh.md: 4564c5376f3e6f9de4d50dc13f2d1f01e6230d91

.agents/notes/implemented/architecture/2026-07-24-separate-context-injection-from-turn-execution.md

@@ -12,7 +12,7 @@ Atomic attachment to an inbox message forced the loop to preserve context throug
 
 Idle `inject()` exposed a second mismatch. Injection did not request model execution, yet the implementation opened and closed a zero-step `injection` turn solely to satisfy the turn-enclosure invariant and obtain a durability checkpoint. A turn therefore sometimes meant “run the agent loop” and sometimes meant “persist context without running it.”
 
-`HookContext` also named its producer rather than its role. The value could come from a native plugin, a hook bridge, prompt admission, or tool post-processing; its stable meaning was additional model-facing context with provenance.
+`HookContext` also named its producer rather than its role. The value could come from a native plugin, a hook bridge, prompt admission, or tool post-processing; its stable meaning was additional model-facing context whose source named the producer.
 
 ## Decision
 
@@ -22,7 +22,7 @@ A caller that owns context delivers an identified, frozen `UserMessage` through
 
 An entering pre-step returns the complete `PreStepDecision.messages` batch for the request being finalized. Tool extension points still return `additionalContexts`, which enter the next-step inbox only after the corresponding tool results. These values are extension-point outputs, not attachments captured from a caller's inbox item.
 
-Every additional context is an independent `UserMessage` whose `source` records provenance. Inbox insertion is durable immediately; admission later records the same value as `user/message`. There is no `context/message`, prompt-prefix placement, stable request delimiter, or prompt envelope. Transcript and UI consumers distinguish direct user messages from injected context by `source`.
+Every additional context is an independent `UserMessage` whose `source` names its producer and carries producer-specific fields. Inbox insertion is durable immediately; admission later records the same value as `user/message`. There is no `context/message`, prompt-prefix placement, stable request delimiter, or prompt envelope. Transcript and UI consumers distinguish direct user messages from injected context by `source`.
 
 ## Injection lifecycle
 
@@ -36,7 +36,7 @@ The loop appends injected `user/message` events only from entered batches inside
 
 ## Extension and caller semantics
 
-The enter branch's `PreStepDecision.messages` is the complete batch for the proposed step. A waterfall listener that delegates with `next()` preserves downstream messages unless it intentionally replaces them; additions follow natural waterfall return order. Tool-result `additionalContexts` retain FIFO order and individual provenance.
+The enter branch's `PreStepDecision.messages` is the complete batch for the proposed step. A waterfall listener that delegates with `next()` preserves downstream messages unless it intentionally replaces them; additions follow natural waterfall return order. Tool-result `additionalContexts` retain FIFO order and each message's source.
 
 Caller-driven injection and current-step context deliberately use different timing. `inject()` joins the next pre-step available and cannot promise that a request already being finalized will consume it. A listener that must affect that exact request returns the context in `PreStepDecision.messages`; downstream rejection or failure then prevents it from materializing.
 

.agents/notes/implemented/architecture/2026-07-24-separate-context-injection-from-turn-execution.zh.md

@@ -12,7 +12,7 @@ agent API 曾用三种相互重叠的方式表示面向模型的补充输入：
 
 空闲状态下的 `inject()` 还暴露了另一处语义错位。注入当时并不请求模型执行，但实现仅为了满足轮次封闭不变量并获得持久性检查点，就会打开并关闭一个零步骤的 `injection` 轮次。于是，当时的轮次有时表示「运行 agent loop」，有时却表示「不运行 agent，仅持久化上下文」。
 
-`HookContext` 的名字也描述了生产方，而非该值的职责。它可能来自原生插件、hook bridge、提示词准入或工具后处理；其稳定含义是带来源信息的额外模型上下文。
+`HookContext` 的名字也描述了生产方，而非该值的职责。它可能来自原生插件、hook bridge、提示词准入或工具后处理；其稳定含义是额外的模型可见上下文，并且 source 会指明生产方。
 
 ## 决策
 
@@ -22,7 +22,7 @@ agent API 曾用三种相互重叠的方式表示面向模型的补充输入：
 
 返回 enter 的 pre-step 会为正在最终确定的请求返回完整的 `PreStepDecision.messages` 批次。工具扩展点仍可返回 `additionalContexts`，这些上下文只会在对应工具结果之后进入 next-step inbox。这些值是扩展点的输出，而不是从调用方 inbox 条目捕获的附件。
 
-每项额外上下文都是独立的 `UserMessage`，并由 `source` 记录来源。inbox 插入会立即持久化；后续准入会将同一个值记录为 `user/message`。不再有 `context/message`、prompt-prefix 放置方式、稳定请求分隔符或提示词封套。transcript 与 UI 消费方通过 `source` 区分直接用户消息和注入上下文。
+每项额外上下文都是独立的 `UserMessage`，其 `source` 会指明生产方，并携带生产方专用字段。inbox 插入会立即持久化；后续准入会将同一个值记录为 `user/message`。不再有 `context/message`、prompt-prefix 放置方式、稳定请求分隔符或提示词封套。transcript 与 UI 消费方通过 `source` 区分直接用户消息和注入上下文。
 
 ## 注入生命周期
 
@@ -36,7 +36,7 @@ loop 只会在轮次内从进入步骤的批次追加注入的 `user/message`。
 
 ## 扩展点与调用方语义
 
-enter 分支的 `PreStepDecision.messages` 是拟议步骤的完整批次。waterfall（瀑布式事件）监听器调用 `next()` 委托时，会保留下游消息，除非有意替换；新增消息遵循 waterfall 的自然返回顺序。工具结果的 `additionalContexts` 保留 FIFO 顺序及各自来源。
+enter 分支的 `PreStepDecision.messages` 是拟议步骤的完整批次。waterfall（瀑布式事件）监听器调用 `next()` 委托时，会保留下游消息，除非有意替换；新增消息遵循 waterfall 的自然返回顺序。工具结果的 `additionalContexts` 保留 FIFO 顺序及每条消息的 source。
 
 调用方主动注入与当前步骤上下文刻意采用不同的时序。`inject()` 会加入下一个可用 pre-step，无法保证正在最终确定的请求会消费它。必须影响该请求的监听器在 `PreStepDecision.messages` 中返回上下文；下游 reject 或失败时，该上下文不会落入日志。
 

.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.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 .agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.md
-2026-07-26-packed-chunk-rows-by-default.md: d6a044676604e4a4512a7a6674edb80e120b2f3c
-2026-07-26-packed-chunk-rows-by-default.zh.md: 0e3e06e0d4185603cd21939589855a51797560a4
+2026-07-26-packed-chunk-rows-by-default.md: 141c9a32a07b5cb4885a21b419df30d45dc1061b
+2026-07-26-packed-chunk-rows-by-default.zh.md: a03654ec82755cadeea734d277c0893db546a5c1

.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.md

@@ -48,7 +48,7 @@ JSONL persistence tests prove that omission writes a packed row, explicit `false
 
 **Remove `packChunks` and always pack.** One writer is simpler, but one-event-per-line output remains useful for diagnostics and for focused mixed-layout compatibility tests. The explicit opt-out preserves those current consumers without weakening the default.
 
-**Batch chunks as logical session events.** This reduces event count, but it delays or reshapes live delivery, renumbers provenance, and requires every UI and replay consumer to understand another streaming unit. Physical packing obtains the storage benefit behind the existing persistence interface.
+**Batch chunks as logical session events.** This reduces event count, but it delays or reshapes live delivery, renumbers the chunk seqs cited by assistant messages, and requires every UI and replay consumer to understand another streaming unit. Physical packing obtains the storage benefit behind the existing persistence interface.
 
 **Keep the branch migrator permanently.** The read-only canonicalizer and snapshot gate own continuing enforcement. A mutation command has value only while in-flight branches still carry the former fixture layout, so its lifetime is explicitly bounded by the removal proposal.
 

.agents/notes/implemented/architecture/2026-07-26-packed-chunk-rows-by-default.zh.md

@@ -48,7 +48,7 @@ JSONL 持久化测试证明：省略选项时会写入打包行，显式传入 `
 
 **删除 `packChunks` 并始终打包。** 只保留一个写入器更简单，但每个事件一行的输出仍适用于诊断和聚焦的混合布局兼容性测试。显式停用选项在不削弱默认值的同时，保留了这些现有消费方。
 
-**把分片批量合并为逻辑会话事件。** 这会减少事件数量，但也会延迟或重塑实时传递，改变溯源信息所引用的序号，并要求每个 UI 和回放消费方理解另一种流式单位。物理打包通过现有持久化接口获得存储收益。
+**把分片批量合并为逻辑会话事件。** 这会减少事件数量，但也会延迟或重塑实时传递，重新编号助手消息引用的分片 seq，并要求每个 UI 和回放消费方理解另一种流式单位。物理打包通过现有持久化接口获得存储收益。
 
 **永久保留分支迁移器。** 只读的规范布局转换器与快照门禁负责持续强制执行。只有在途分支仍携带旧 fixture 布局时，会修改仓库内容的命令才有价值，因此移除提案明确限定了其生命周期。
 

.agents/notes/implemented/architecture/2026-07-28-identified-immutable-message-values.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 .agents/notes/implemented/architecture/2026-07-28-identified-immutable-message-values.md
-2026-07-28-identified-immutable-message-values.md: 472de8b133ba323c3e1ff5e53c8dacb3d66525c5
-2026-07-28-identified-immutable-message-values.zh.md: d322ff93765294478d2adce41a5903f3a7fdd67d
+2026-07-28-identified-immutable-message-values.md: f1e0e8c0b42bd2dc4b3c729dc15b5f2a36b98338
+2026-07-28-identified-immutable-message-values.zh.md: 0d1f2fcf5be76089f1c135757a7a50729f099ccd

.agents/notes/implemented/architecture/2026-07-28-identified-immutable-message-values.md

@@ -8,13 +8,13 @@ English | [中文](2026-07-28-identified-immutable-message-values.zh.md)
 
 The harness had several message-shaped representations with different identity rules. Agent input acquired an inbox correlation id only when the loop accepted it, while durable user messages, assistant messages, tool results, and model-request messages could have no identity. Prompt admission therefore sat between creation and identity, and equivalent content was copied across live events, durable events, and model requests without one value that named the message throughout its lifetime.
 
-This made identity a routing side effect rather than a message invariant. Producers could not refer to a message before calling the agent, prompt hooks received content and source separately, and later projections had to reconstruct a message while deciding whether an id existed. Immutability also began at different boundaries: some inputs were frozen by the loop, some only by session append, and provider-produced assistant output used a separate provenance-bearing shape.
+This made identity a routing side effect rather than a message invariant. Producers could not refer to a message before calling the agent, prompt hooks received content and source separately, and later projections had to reconstruct a message while deciding whether an id existed. Immutability also began at different boundaries: some inputs were frozen by the loop, some only by session append, and provider-produced assistant output used a separate shape carrying provider, model, and replay state.
 
 ## Decision
 
 `@deepseek-ai/dsh-llm` owns one `Message` value with required `id`, `role`, `content`, and `source`. `MessageId` is opaque and shared by user, assistant, and tool-result messages. A message receives its id at creation, before inbox routing, claim, pre-step rewriting, durable append, or request projection. The same id survives every representation boundary.
 
-`createMessage(input)` is the canonical role-generic creation boundary. It mints a `MessageId`, detaches the supplied role, content, and source, and deep-freezes the complete value before returning it. `createUserMessage({ content, source })` fixes the user role for prompt and context producers. `createAssistantMessage({ content, source })` fixes both the assistant role and the model source kind, so model-output producers supply only content and model provenance. All creation helpers exclude an input id so callers cannot accidentally present creation as import. `freezeMessage(message)` is the separate import or transformation boundary: it detaches and deep-freezes a message whose identity already exists, without minting a replacement.
+`createMessage(input)` is the canonical role-generic creation boundary. It mints a `MessageId`, detaches the supplied role, content, and source, and deep-freezes the complete value before returning it. `createUserMessage({ content, source })` fixes the user role for prompt and context producers. `createAssistantMessage({ content, source })` fixes both the assistant role and the model source kind, so model-output producers supply only content plus provider, model, and optional replay state. All creation helpers exclude an input id so callers cannot accidentally present creation as import. `freezeMessage(message)` is the separate import or transformation boundary: it detaches and deep-freezes a message whose identity already exists, without minting a replacement.
 
 The helpers live in `dsh-llm` beside the base message vocabulary because their complete contracts depend only on that vocabulary. `createToolResultMessage()` belongs with the other creation helpers: it couples a tool call id to the exact user-role tool-result block and source without depending on session state or events. `dsh-session` consumes complete messages rather than owning their construction.
 
@@ -40,7 +40,7 @@ Every message producer must choose creation or import explicitly, and tests cons
 
 Live inbox events, durable events, derived history, and model requests can correlate one message without content equality or envelope-specific ids. Pending-input policy and UI attachment cleanup can compare `MessageId` before a turn exists, while claims retain that identity inside the open turn. Deep freezing prevents a producer, hook, or observer from changing the value after identity is established.
 
-The shared representation removes the old `UserMessageData`/`AgentMessage` split and folds provider provenance into typed message sources. Event envelopes still own facts that are not message semantics, such as turn and step position, token usage, internal tool failure identity, and presentation metadata.
+The shared representation removes the old `UserMessageData`/`AgentMessage` split and puts provider, model, and optional replay state into typed message sources. Event envelopes still own facts that are not message semantics, such as turn and step position, token usage, internal tool failure identity, and presentation metadata.
 
 The message and helper unit tests pin immediate identity, detachment, deep immutability, and preservation of an imported id. Agent-loop tests pin identity across admission, inbox lifecycle, durable append, content rewriting, and cancellation; session tests pin frozen derivation and identity-preserving replacement.
 

.agents/notes/implemented/architecture/2026-07-28-identified-immutable-message-values.zh.md

@@ -8,13 +8,13 @@ Status: implemented
 
 harness 曾存在多种形似消息的表示，各自采用不同的标识规则。agent（智能体）输入只有在 agent loop 接受后才会取得 inbox 关联 id，而持久用户消息、assistant 消息、工具结果和模型请求消息都可能没有标识。因此，提示词准入介于创建消息与建立标识之间；等价内容会在实时事件、持久事件和模型请求之间复制，却没有一个值能在消息的整个生命周期中标识它。
 
-这使标识成为路由的副作用，而不是消息不变量。生产方无法在调用 agent 前引用一条消息，提示词钩子会分别接收内容和来源，后续投影则必须一边重建消息，一边决定 id 是否存在。不可变性也从不同边界开始：部分输入由 agent loop 冻结，部分直到会话追加时才冻结，由提供方生成的 assistant 输出则使用另一种携带溯源信息的形状。
+这使标识成为路由的副作用，而不是消息不变量。生产方无法在调用 agent 前引用一条消息，提示词钩子会分别接收内容和来源，后续投影则必须一边重建消息，一边决定 id 是否存在。不可变性也从不同边界开始：部分输入由 agent loop 冻结，部分直到会话追加时才冻结，由提供方生成的 assistant 输出则使用另一种携带提供方、模型和回放状态的形状。
 
 ## 决策
 
 `@deepseek-ai/dsh-llm` 持有唯一一种 `Message` 值，其 `id`、`role`、`content` 和 `source` 均为必填。`MessageId` 是不透明标识，由用户消息、assistant 消息和工具结果消息共享。消息在创建时就会获得 id，早于 inbox 路由、领取、pre-step 改写、持久追加或请求投影。同一个 id 会跨越每个表示边界。
 
-`createMessage(input)` 是角色通用的规范创建边界。它会生成 `MessageId`，将传入的角色、内容和来源与调用方对象解除引用关系，并在返回完整值前将其深度冻结。`createUserMessage({ content, source })` 为提示词和上下文生产方固定 user 角色。`createAssistantMessage({ content, source })` 同时固定 assistant 角色与模型来源类别，因此模型输出生产方只需提供内容和模型溯源信息。所有创建辅助函数的输入都不包含 id，因此调用方不会意外地把新消息的创建伪装成已有消息的导入。`freezeMessage(message)` 是独立的导入或转换边界：它会将已有标识的消息与调用方对象解除引用关系并深度冻结，不会生成替代标识。
+`createMessage(input)` 是角色通用的规范创建边界。它会生成 `MessageId`，将传入的角色、内容和来源与调用方对象解除引用关系，并在返回完整值前将其深度冻结。`createUserMessage({ content, source })` 为提示词和上下文生产方固定 user 角色。`createAssistantMessage({ content, source })` 同时固定 assistant 角色与模型来源类别，因此模型输出生产方只需提供内容，以及提供方、模型和可选的回放状态。所有创建辅助函数的输入都不包含 id，因此调用方不会意外地把新消息的创建伪装成已有消息的导入。`freezeMessage(message)` 是独立的导入或转换边界：它会将已有标识的消息与调用方对象解除引用关系并深度冻结，不会生成替代标识。
 
 这些辅助函数位于基础消息词汇旁的 `dsh-llm` 中，因为它们的完整约定只依赖该词汇。`createToolResultMessage()` 与其他创建辅助函数同属此处：它使用同一个工具调用 id，将工具来源与确切的 user-role 工具结果块耦合起来，不依赖会话状态或事件。`dsh-session` 只消费完整消息，不负责构造它们。
 
@@ -40,7 +40,7 @@ harness 曾存在多种形似消息的表示，各自采用不同的标识规则
 
 实时 inbox 事件、持久事件、派生历史和模型请求可以关联同一条消息，无需比较内容或使用封装专用 id。待处理输入策略和 UI 附件清理可以在轮次存在之前比较 `MessageId`，领取后则会在已打开的轮次内保留该标识。深度冻结可以防止生产方、钩子或观察方在标识建立后更改消息值。
 
-共享表示移除了旧的 `UserMessageData`/`AgentMessage` 划分，并将提供方溯源信息纳入带类型的消息来源。事件封装仍持有不属于消息语义的事实，例如轮次与步骤位置、token 用量、内部工具失败标识和呈现元数据。
+共享表示移除了旧的 `UserMessageData`/`AgentMessage` 划分，并将提供方、模型和可选的回放状态放入带类型的消息来源。事件封装仍持有不属于消息语义的事实，例如轮次与步骤位置、token 用量、内部工具失败标识和呈现元数据。
 
 消息和辅助函数的单元测试会锁定即时标识、解除输入引用、深度不可变性，以及导入 id 的保留。agent loop 测试会锁定标识跨越准入、inbox 生命周期、持久追加、内容改写和取消的行为；会话测试会锁定冻结派生和保留标识的替换行为。
 

.agents/notes/implemented/architecture/2026-07-29-terminal-llm-stream-failures.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 .agents/notes/implemented/architecture/2026-07-29-terminal-llm-stream-failures.md
-2026-07-29-terminal-llm-stream-failures.md: 1e26973360f07c212016c6a44103448a3510a75b
-2026-07-29-terminal-llm-stream-failures.zh.md: d3eeb0534f6cb8d4d1ad167cca089314eb02b513
+2026-07-29-terminal-llm-stream-failures.md: 3f9fc4255e5b2462e14f73414894c5912191c595
+2026-07-29-terminal-llm-stream-failures.zh.md: a73edd89180b22225eb97804857bd1d3b93a4c4a

.agents/notes/implemented/architecture/2026-07-29-terminal-llm-stream-failures.md

@@ -28,10 +28,10 @@ The agent loop consumes one failure representation. It iterates and logs chunks
 
 **Require every adapter to emit failure chunks and forbid throws.** Library iterators, transports, and JavaScript dispatch can still throw. Requiring every adapter to reproduce the same catch boundary duplicates ownership and does not protect a direct `LlmService` consumer from an incomplete implementation.
 
-**Catch every iteration error in the agent loop.** The loop cannot reliably distinguish provider failure from middleware, session append, cancellation, or assembly failure without restoring the same sidecar provenance mechanism. Classification belongs where the adapter call is made.
+**Catch every iteration error in the agent loop.** The loop cannot reliably distinguish provider failure from middleware, session append, cancellation, or assembly failure without restoring a sidecar map from stream objects to the adapter calls that created them. Classification belongs where the adapter call is made.
 
 **Return a `Result` before streaming.** A pre-stream result cannot represent a transport failure after partial output without adding a second response lifecycle. The existing terminal chunk already represents both early and late attempt outcomes.
 
 ## Consequences
 
-All `LlmService.stream()` consumers receive adapter operational failures through one typed terminal protocol, while programming and lifecycle failures retain ordinary exception semantics. Recovery gives up exact thrown-object identity and exposes only detached provider-neutral facts. The stream service owns slightly more adapter plumbing, but consumers delete provenance catches and stream-keyed metadata. Prepared calls carry their policy explicitly, and middleware-only routing remains visibly policy-free.
+All `LlmService.stream()` consumers receive adapter operational failures through one typed terminal protocol, while programming and lifecycle failures retain ordinary exception semantics. Recovery gives up exact thrown-object identity and exposes only detached provider-neutral facts. The stream service owns slightly more adapter plumbing, but consumers delete catches that identify which adapter threw and delete stream-keyed metadata. Prepared calls carry their policy explicitly, and middleware-only routing remains visibly policy-free.

.agents/notes/implemented/architecture/2026-07-29-terminal-llm-stream-failures.zh.md

@@ -28,10 +28,10 @@ agent loop 只消费一种失败表示。它不再使用分类 catch，而是直
 
 **要求所有适配器发出失败 chunk，并禁止抛出。** 库 iterator、transport 与 JavaScript 分发仍可能抛错。要求每个适配器复制同一 catch 边界会重复归属，也无法保护 `LlmService` 的直接消费方免受不完整实现影响。
 
-**在 agent loop 中捕获所有迭代错误。** 如果不恢复同一套 sidecar 溯源机制，loop 无法可靠区分提供方失败与 middleware、session append、取消或组装失败。分类属于发起适配器调用的边界。
+**在 agent loop 中捕获所有迭代错误。** 如果不恢复一张从流对象到创建该对象的适配器调用的 sidecar 映射，loop 无法可靠区分提供方失败与 middleware、session append、取消或组装失败。分类属于发起适配器调用的边界。
 
 **在流式输出前返回 `Result`。** 流前结果无法表示部分输出之后的传输失败，除非增加第二套响应生命周期。现有终止 chunk 已能表示早期和后期尝试结果。
 
 ## Consequences
 
-所有 `LlmService.stream()` 消费方都通过一种带类型的终止协议接收适配器运行失败，而编程与生命周期失败保留普通异常语义。恢复放弃精确抛出对象身份，只暴露与原对象分离的提供方无关事实。流服务承担略多的适配器管道工作，但消费方删除了溯源 catch 与以 stream 为 key 的元数据。准备完成的调用显式携带策略，而仅由 middleware 路由的调用仍明确没有策略。
+所有 `LlmService.stream()` 消费方都通过一种带类型的终止协议接收适配器运行失败，而编程与生命周期失败保留普通异常语义。恢复放弃精确抛出对象身份，只暴露与原对象分离的提供方无关事实。流服务承担略多的适配器管道工作，但消费方删除了用于识别哪个适配器抛出异常的 catch，也删除了以 stream 为 key 的元数据。准备完成的调用显式携带策略，而仅由 middleware 路由的调用仍明确没有策略。

.agents/notes/implemented/architecture/2026-07-30-adapter-owned-max-token-defaults.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 .agents/notes/implemented/architecture/2026-07-30-adapter-owned-max-token-defaults.md
-2026-07-30-adapter-owned-max-token-defaults.md: a522848fd4482f84859e587505b6a5e6f5c72d60
-2026-07-30-adapter-owned-max-token-defaults.zh.md: 8db6a06199fc1c4e73c86492d12dc86edafe8c7e
+2026-07-30-adapter-owned-max-token-defaults.md: c6a784f23a028f500c6e6ff80dc3e13eee93ec1b
+2026-07-30-adapter-owned-max-token-defaults.zh.md: 3caa81126c5d45db7ef5c84540b0deb81764aaf4

.agents/notes/implemented/architecture/2026-07-30-adapter-owned-max-token-defaults.md

@@ -12,7 +12,7 @@ An LLM adapter could serialize an explicit `GenerateOptions.maxTokens`, but its
 
 `LlmResolvedModelInfo.defaultMaxTokens` carries an optional adapter-configured per-request output cap for one exact provider/model route. `LlmService` validates it as a positive safe integer and materializes it into `LlmCallConfig.maxTokens` only when the caller omitted a value. A prepared call identifies materialized `maxTokens` and `reasoningEffort` fields as adapter defaults; explicit request or Agent options remain unmarked and therefore win without clamping.
 
-The agent loop continues to prepare calls before logging `request/header`, so the effective config and its adapter-default provenance become durable request facts before dispatch. Before the next `agent/request` waterfall, the loop removes marked fields from the proposal; exact-model resolution then materializes the current route's defaults again. A provider/model switch therefore cannot mistake a previous adapter's default for an explicit override, while explicit conversation values persist. Direct `LlmService.stream()` calls resolve the same default at the final adapter boundary. The field is a request default rather than a hard model output limit; adapters that preserve provider-owned defaults omit it.
+The agent loop continues to prepare calls before logging `request/header`, so the effective config and markers for fields supplied by adapter defaults become durable request facts before dispatch. Before the next `agent/request` waterfall, the loop removes marked fields from the proposal; exact-model resolution then materializes the current route's defaults again. A provider/model switch therefore cannot mistake a previous adapter's default for an explicit override, while explicit conversation values persist. Direct `LlmService.stream()` calls resolve the same default at the final adapter boundary. The field is a request default rather than a hard model output limit; adapters that preserve provider-owned defaults omit it.
 
 The native DeepSeek adapter exposes `maxTokens` in Cordis config with a 256,000-token default and maps the effective value to `max_tokens`. Its default context capacity is 1,000,000 tokens: both built-in V4 entries publish that exact capacity, while configured entries without capacity and unlisted pass-through ids inherit the same adapter-wide fallback.
 
@@ -28,6 +28,6 @@ The native DeepSeek adapter exposes `maxTokens` in Cordis config with a 256,000-
 
 ## Consequences
 
-DeepSeek conversations send `max_tokens: 256000` by default, and the same value plus its adapter provenance appear in the session request header. Deployments can change the adapter default through `llm-deepseek.config.maxTokens`; per-agent and per-request values override it. Changing the route rematerializes the new exact adapter's default instead of carrying DeepSeek's derived value forward. Other adapters retain their existing behavior until they intentionally publish `defaultMaxTokens`.
+DeepSeek conversations send `max_tokens: 256000` by default, and the session request header records both the value and that the adapter supplied it. Deployments can change the adapter default through `llm-deepseek.config.maxTokens`; per-agent and per-request values override it. Changing the route rematerializes the new exact adapter's default instead of carrying DeepSeek's derived value forward. Other adapters retain their existing behavior until they intentionally publish `defaultMaxTokens`.
 
 The 256,000-token output budget reserves a large part of the one-million-token context on endpoints that pre-allocate requested output. Deployments whose gateway or model supports a smaller budget must lower `maxTokens`; the explicit configuration is preferable to an undocumented provider fallback.

.agents/notes/implemented/architecture/2026-07-30-adapter-owned-max-token-defaults.zh.md

@@ -12,7 +12,7 @@ LLM（大语言模型）适配器可以序列化显式的 `GenerateOptions.maxTo
 
 `LlmResolvedModelInfo.defaultMaxTokens` 携带一条确切提供方／模型路由的可选单次请求输出上限，该值由适配器配置。`LlmService` 将其校验为正安全整数，并且仅在调用方省略值时才填入 `LlmCallConfig.maxTokens`。准备后的调用会将已填入的 `maxTokens` 和 `reasoningEffort` 字段标记为适配器默认值；显式请求值或 Agent 选项不带该标记，因此优先且不会被自动调整。
 
-agent loop 仍在记录 `request/header` 前准备调用，因此生效配置及其适配器默认值来源会在分派前成为持久请求事实。下一次 `agent/request` waterfall（瀑布式事件）前，agent loop 会从提议中移除带标记字段，随后精确模型解析会再次填入当前路由的默认值。因此，切换提供方／模型不会把前一个适配器的默认值误当成显式覆盖，而显式对话值则会保留。直接调用 `LlmService.stream()` 时，也会在最终适配器边界解析同一默认值。该字段是请求默认值，而非模型输出硬上限；保留提供方持有默认值的适配器会省略它。
+agent loop 仍在记录 `request/header` 前准备调用，因此生效配置和标明哪些字段由适配器默认值填入的标记，会在分派前成为持久请求事实。下一次 `agent/request` waterfall（瀑布式事件）前，agent loop 会从提议中移除带标记字段，随后精确模型解析会再次填入当前路由的默认值。因此，切换提供方／模型不会把前一个适配器的默认值误当成显式覆盖，而显式对话值则会保留。直接调用 `LlmService.stream()` 时，也会在最终适配器边界解析同一默认值。该字段是请求默认值，而非模型输出硬上限；保留提供方持有默认值的适配器会省略它。
 
 原生 DeepSeek 适配器在 Cordis 配置中公开 `maxTokens`，默认值为 256,000 token，并将生效值映射为 `max_tokens`。其默认上下文容量为 1,000,000 token：两个内置 V4 配置项均公布这一精确容量；不含容量的已配置项和未列出的原样传递 id 则继承同一个适配器级回退值。
 
@@ -28,6 +28,6 @@ agent loop 仍在记录 `request/header` 前准备调用，因此生效配置及
 
 ## Consequences
 
-DeepSeek 对话默认发送 `max_tokens: 256000`，会话请求 header 中也会出现相同的值及其适配器来源。部署可以通过 `llm-deepseek.config.maxTokens` 更改适配器默认值；每个 agent 和每次请求的值都会覆盖它。更改路由会重新填入新的精确适配器默认值，而不是继续沿用 DeepSeek 派生出的值。其他适配器会保留现有行为，直至主动公布 `defaultMaxTokens`。
+DeepSeek 对话默认发送 `max_tokens: 256000`，会话请求 header 会记录该值，并记录该值由适配器提供。部署可以通过 `llm-deepseek.config.maxTokens` 更改适配器默认值；每个 agent 和每次请求的值都会覆盖它。更改路由会重新填入新的精确适配器默认值，而不是继续沿用 DeepSeek 派生出的值。其他适配器会保留现有行为，直至主动公布 `defaultMaxTokens`。
 
 对于预分配请求输出的端点，256,000 token 的输出预算会占用 1,000,000 token 上下文中的很大部分。如果部署使用的 gateway 或模型仅支持较小预算，则必须调低 `maxTokens`；显式配置优于未记录的提供方回退值。

.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.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 .agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.md
-2026-08-08-bounded-session-persistence-write-batching.md: 46dc612492fa1bfa805f77f865f14b168f52776f
-2026-08-08-bounded-session-persistence-write-batching.zh.md: fe1227f7706ba67a12bd7f19db5c0c57cfcfca4f
+2026-08-08-bounded-session-persistence-write-batching.md: 6d77ef90276dcf143bb63019f801379b702b43b2
+2026-08-08-bounded-session-persistence-write-batching.zh.md: 37ebb1566f1826ebbc6fcaf3dca11b65aed3a8ba

.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.md

@@ -8,7 +8,7 @@ English | [中文](2026-08-08-bounded-session-persistence-write-batching.zh.md)
 
 Streaming responses can emit many `assistant/chunk` events in a short interval. The persistence coordinator previously scheduled a backend append as soon as an idle queue received one event. Events arriving while that append was active shared a follow-up batch, but a fast backend could still produce many small durable appends. Each JSONL append creates and syncs a Zstandard frame or raw suffix, while each SQLite append opens and commits a transaction and increments the session revision.
 
-Dropping chunk events or replacing them with assembled messages would reduce logical storage, but it would also change the event log, replay, sequence numbers, timestamps, and provenance. The write-amplification problem does not require that larger semantic change.
+Dropping chunk events or replacing them with assembled messages would reduce logical storage, but it would also change the event log, replay, sequence numbers, timestamps, and the chunk seqs cited by assistant messages. The write-amplification problem does not require that larger semantic change.
 
 ### Quantified baseline
 
@@ -36,7 +36,7 @@ This decision supersedes only the immediate scheduling cadence in [Collapse live
 
 ## Alternatives considered
 
-**Do not persist streaming chunk events.** Rejected here: it changes the event-sourced authority and recovery semantics rather than only physical write cadence. The existing [assembled-message rejection](../../rejected/simplification/2026-06-20-assembled-assistant-messages-only.md) remains the guardrail until a no-information-loss replacement defines replay, fork, provenance, sequence, and crash behavior independently. The [packed-row decision](2026-07-26-packed-chunk-rows-by-default.md) remains the complementary JSONL storage-size optimization.
+**Do not persist streaming chunk events.** Rejected here: it changes the event-sourced authority and recovery semantics rather than only physical write cadence. The existing [assembled-message rejection](../../rejected/simplification/2026-06-20-assembled-assistant-messages-only.md) remains the guardrail until a no-information-loss replacement defines replay, fork, cited source-event links, sequence, and crash behavior independently. The [packed-row decision](2026-07-26-packed-chunk-rows-by-default.md) remains the complementary JSONL storage-size optimization.
 
 **Write only at semantic checkpoints.** Rejected: it maximizes batching but makes the ordinary crash-loss window depend on a separately mounted policy. Bounded background writes preserve progress between checkpoints while mandatory flushes keep their stronger ordering contract.
 

.agents/notes/implemented/architecture/2026-08-08-bounded-session-persistence-write-batching.zh.md

@@ -8,7 +8,7 @@ Status: implemented
 
 流式响应可能会在短时间内发出大量 `assistant/chunk` 事件。此前，只要空闲队列收到一个事件，持久化协调器就会立即调度一次后端追加。该追加仍在进行时到达的事件会共用一个后续批次，但如果后端速度很快，仍可能产生大量小规模的持久化追加。每次 JSONL 追加都会创建并同步一个 Zstandard 帧或原始格式后缀，而每次 SQLite 追加都会打开并提交一个事务，同时递增会话修订版本。
 
-丢弃分片事件或用组装后的消息替代它们可以减少逻辑存储量，但也会改变事件日志、回放、序列号、时间戳和来源信息。写放大问题不要求采取这项语义变化更大的方案。
+丢弃分片事件或用组装后的消息替代它们可以减少逻辑存储量，但也会改变事件日志、回放、序列号、时间戳，以及助手消息引用的分片 seq。写放大问题不要求采取这项语义变化更大的方案。
 
 ### 量化基线
 
@@ -36,7 +36,7 @@ SQLite 每个逻辑事件存储一行，因此同样的逻辑日志会分别保
 
 ## 备选方案
 
-**不持久化流式分片事件。** 这里不采纳：这会改变事件日志作为真源的地位及恢复语义，而不只是改变物理写入节奏。在无信息损失的替代方案独立定义回放、fork、来源信息、序列和崩溃行为之前，现有的[拒绝仅保留组装消息的决策](../../rejected/simplification/2026-06-20-assembled-assistant-messages-only.md)仍是防护规则。[打包行决策](2026-07-26-packed-chunk-rows-by-default.md)仍是配套的 JSONL 存储体积优化。
+**不持久化流式分片事件。** 这里不采纳：这会改变事件日志作为真源的地位及恢复语义，而不只是改变物理写入节奏。在无信息损失的替代方案独立定义回放、fork、引用源事件的关联、序列和崩溃行为之前，现有的[拒绝仅保留组装消息的决策](../../rejected/simplification/2026-06-20-assembled-assistant-messages-only.md)仍是防护规则。[打包行决策](2026-07-26-packed-chunk-rows-by-default.md)仍是配套的 JSONL 存储体积优化。
 
 **仅在语义检查点写入。** 不采纳：此方案会最大化批处理，却让普通的崩溃丢失窗口取决于另行挂载的策略。有界后台写入会在检查点之间持久化进度，而强制 flush 继续提供更强的顺序约定。