From ea47c3280504c21ba795e2d20b514add43251751 Mon Sep 17 00:00:00 2001
From: Hypatia May <hypatiamay@outlook.com>
Date: Mon, 10 Aug 2026 16:09:04 +0800
Subject: [PATCH] review: one fold implementation, uniform end-edge absence,
 partial text in tool errors, snapshot scenario
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Address ds-review-bot on #2127:

- assistant-output: the rule has ONE implementation, the incremental
  AssistantOutputFold (push/pushText/collect); finalAssistantOutput folds a
  complete suffix, the SDK backend folds notification events, and the ACP
  backend folds raw chunk text into the same streamed fallback.
- subagent/end.lastAssistantMessage: 'no output' is encoded once — absent,
  never [], on both lifecycle shapes (observeRun now omits empty output).
- tool-subagent: a non-completed foreground result stays isError but appends
  the child's preserved partial text after the stop-reason headline.
- Authored keyless snapshot scenario subagent-max-tokens-partial pins the
  assembled transcript: the child's committed log carries the usage-only
  empty message and the parent's tool result carries the partial answer.
- Rule-boundary sentence (message wins over later streamed text) and the
  consumer half recorded in the Agent Note; comments trimmed to pointers.
---
 ...nt-empty-terminal-message-output.i18n.yaml |  4 +-
 ...-subagent-empty-terminal-message-output.md |  8 +-
 ...bagent-empty-terminal-message-output.zh.md |  8 +-
 examples/acp-agent/tests/acp.snapshot.ts      |  7 ++
 .../subagent-max-tokens-partial/input.json    | 14 +++
 .../session.1.jsonl                           | 30 +++++++
 .../subagent-max-tokens-partial/session.jsonl | 26 ++++++
 .../stdout.expected.jsonl                     |  4 +
 packages/subagent/subagent-acp/src/run.ts     | 18 ++--
 packages/subagent/subagent-dsh-sdk/src/run.ts | 26 ++----
 .../subagent/subagent-inprocess/src/index.ts  |  4 +-
 packages/subagent/subagent/README.i18n.yaml   |  4 +-
 packages/subagent/subagent/README.md          |  2 +-
 packages/subagent/subagent/README.zh.md       |  2 +-
 .../subagent/subagent/src/assistant-output.ts | 86 ++++++++++++-------
 packages/subagent/subagent/src/index.ts       |  2 +-
 packages/subagent/subagent/src/lifecycle.ts   |  4 +-
 .../subagent/tests/assistant-output.spec.ts   | 25 +++---
 .../subagent/subagent/tests/service.spec.ts   | 12 +++
 .../subagent/tool-subagent/README.i18n.yaml   |  4 +-
 packages/subagent/tool-subagent/README.md     |  2 +-
 packages/subagent/tool-subagent/README.zh.md  |  2 +-
 packages/subagent/tool-subagent/src/index.ts  | 20 ++++-
 .../tool-subagent/tests/tool-subagent.spec.ts |  3 +
 24 files changed, 221 insertions(+), 96 deletions(-)
 create mode 100644 examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/input.json
 create mode 100644 examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/session.1.jsonl
 create mode 100644 examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/session.jsonl
 create mode 100644 examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/stdout.expected.jsonl

diff --git a/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.i18n.yaml
index 9ec2d8bb33..537cb88062 100644
--- a/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.i18n.yaml
+++ b/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.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/bug-fix/2026-08-10-subagent-empty-terminal-message-output.md
-2026-08-10-subagent-empty-terminal-message-output.md: ece2a55930aabfaeaf88ef1757918296b32bfea5
-2026-08-10-subagent-empty-terminal-message-output.zh.md: 00b8f5bd7bd8e71935f68af2c35521f2b2377186
+2026-08-10-subagent-empty-terminal-message-output.md: d90047c07a300a1afbc42c7db1a4fefa25d56764
+2026-08-10-subagent-empty-terminal-message-output.zh.md: 0a5ce02dccef422dc75bc980d104f41f116427f2
diff --git a/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.md b/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.md
index ece2a55930..d90047c07a 100644
--- a/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.md
+++ b/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.md
@@ -10,9 +10,11 @@ The agent loop appends an EMPTY-content `assistant/message` when a `max-tokens`
 
 ## Decision
 
-`dsh-subagent` owns one canonical selection rule in `src/assistant-output.ts`: the last NON-EMPTY assistant message wins; without one, the accumulated `text-delta` stream is the answer; empty-content messages never contribute. `finalAssistantOutput(events)` applies the rule to an event suffix (the in-process `readResult` and the Activation capture), and `assistantMessageOutput(event)` is the same per-event predicate for the SDK backend's incremental fold. The contract is stated once at `SubagentResult.output` and mirrored by the subsystem reference; `subagent/end.lastAssistantMessage` declares it selects by the same rule. A `max-tokens` or `aborted` finish still reports its honest stop reason; only output selection changed.
+`dsh-subagent` owns one canonical selection rule in `src/assistant-output.ts`: the last NON-EMPTY assistant message wins; without one, the accumulated `text-delta` stream is the answer; empty-content messages never contribute. The rule has one implementation, the incremental `AssistantOutputFold` (`push(event)` for session-event transports, `pushText(text)` for chunk-only transports, `collect()` to select), and `finalAssistantOutput(events)` applies it to a complete event suffix (the in-process `readResult` and the Activation capture). The SDK backend folds notification events; the ACP backend, which surfaces no complete assistant messages, folds raw chunk text into the same streamed fallback. The contract is stated once at `SubagentResult.output` and mirrored by the subsystem reference; `subagent/end.lastAssistantMessage` selects by the same rule, and "no output" has one encoding on that edge — the field is absent, never an empty array, on both the one-shot and continuable lifecycle shapes. A `max-tokens` or `aborted` finish still reports its honest stop reason; only output selection changed.
 
-The ACP backend accumulates chunks only and was never affected. The fake SDK runtime gained a `FAKE_EMPTY_MESSAGE` mode so the keyless backend test can script a usage-only terminal message.
+The foreground delegation tool observes the same selection: a non-`completed` result stays an `isError` tool result, but its message appends the child's preserved partial text after the stop-reason headline, so the parent model sees the truncated answer instead of a bare failure.
+
+The fake SDK runtime gained a `FAKE_EMPTY_MESSAGE` mode so the keyless backend test can script a usage-only terminal message, and the authored `subagent-max-tokens-partial` ACP snapshot scenario pins the assembled transcript: a scripted child streams text plus a tool call, is cut off by a tool-only max-tokens step (the empty usage-only message appears in its committed log), and the parent's tool result carries the partial answer.
 
 ## Alternatives considered
 
@@ -24,4 +26,4 @@ The ACP backend accumulates chunks only and was never affected. The fake SDK run
 
 ## Consequences
 
-Multi-step children cut off at max-tokens report their earlier text; cancelled in-process children keep the text streamed before the abort; one-shot and continuable `subagent/end` edges agree with `SubagentResult.output`. A message whose content is non-empty but textless (for example reasoning-only) still wins over streamed text — the rule is about empty content, not text presence. Regression tests in all three packages script the empty-terminal-message and cancel paths and fail under the previous selections.
+Multi-step children cut off at max-tokens report their earlier text; cancelled in-process children keep the text streamed before the abort; one-shot and continuable `subagent/end` edges agree with `SubagentResult.output`. A message whose content is non-empty but textless (for example reasoning-only) still wins over streamed text — the rule is about empty content, not text presence. A non-empty message also wins over text streamed AFTER it: a child cancelled while streaming a later step reports its earlier complete message, matching the SDK backend's documented contract, with the stop reason signalling the truncation. Regression tests in all three packages script the empty-terminal-message and cancel paths and fail under the previous selections.
diff --git a/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.zh.md b/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.zh.md
index 00b8f5bd7b..0a5ce02dcc 100644
--- a/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.zh.md
+++ b/.agents/notes/implemented/bug-fix/2026-08-10-subagent-empty-terminal-message-output.zh.md
@@ -10,9 +10,11 @@ Status: implemented
 
 ## 决策
 
-`dsh-subagent` 在 `src/assistant-output.ts` 中拥有唯一的规范选取规则：最后一条**非空** assistant 消息优先；没有时，累积的 `text-delta` 流就是回答；空内容消息从不参与。`finalAssistantOutput(events)` 把该规则应用于事件后缀（进程内 `readResult` 与 Activation capture），`assistantMessageOutput(event)` 是同一规则的逐事件谓词，供 SDK 后端的增量折叠使用。契约在 `SubagentResult.output` 处声明一次，并由子系统参考文档镜像；`subagent/end.lastAssistantMessage` 声明按同一规则选取。`max-tokens` 或 `aborted` 终止仍然如实上报其终止原因；只有输出选取发生了变化。
+`dsh-subagent` 在 `src/assistant-output.ts` 中拥有唯一的规范选取规则：最后一条**非空** assistant 消息优先；没有时，累积的 `text-delta` 流就是回答；空内容消息从不参与。规则只有一个实现，即增量的 `AssistantOutputFold`（会话事件传输用 `push(event)`，仅分块传输用 `pushText(text)`，`collect()` 完成选取）；`finalAssistantOutput(events)` 把它应用于完整的事件后缀（进程内 `readResult` 与 Activation capture）。SDK 后端折叠通知事件；ACP 后端不产生完整 assistant 消息，因此把原始分块文本折叠进同一个流式兜底。契约在 `SubagentResult.output` 处声明一次，并由子系统参考文档镜像；`subagent/end.lastAssistantMessage` 按同一规则选取，且"无输出"在该边沿只有一种编码——字段缺省，绝不是空数组，一次性与 continuable 两种生命周期形态一致。`max-tokens` 或 `aborted` 终止仍然如实上报其终止原因；只有输出选取发生了变化。
 
-ACP 后端只累积分块，从未受影响。fake SDK runtime 新增 `FAKE_EMPTY_MESSAGE` 模式，使无密钥后端测试能够脚本化一条仅承载 usage 的终止消息。
+前台委派工具观察同一选取结果：非 `completed` 的结果仍是 `isError` 工具结果，但其消息在终止原因标题之后附带子代理保留下来的部分文本，父模型看到的是被截断的回答而不是一句干巴巴的失败。
+
+fake SDK runtime 新增 `FAKE_EMPTY_MESSAGE` 模式，使无密钥后端测试能够脚本化一条仅承载 usage 的终止消息；authored 的 `subagent-max-tokens-partial` ACP snapshot 场景钉住了组装后的 transcript：脚本化的子代理先流式输出文本和一次工具调用，再被仅含工具调用的 max-tokens 步骤截断（空的 usage-only 消息出现在其提交的日志中），父侧工具结果携带部分回答。
 
 ## 考虑过的替代方案
 
@@ -24,4 +26,4 @@ ACP 后端只累积分块，从未受影响。fake SDK runtime 新增 `FAKE_EMPT
 
 ## 后果
 
-被 max-tokens 截断的多步子代理会报告其更早的文本；被取消的进程内子代理保留中止前已流式的文本；一次性与 continuable 的 `subagent/end` 边沿与 `SubagentResult.output` 一致。内容非空但不含文本的消息（例如仅含 reasoning）仍然优先于流式文本——规则针对的是内容为空，而非文本缺失。三个包中的回归测试脚本化了空终止消息与取消路径，并在先前的选取实现下失败。
+被 max-tokens 截断的多步子代理会报告其更早的文本；被取消的进程内子代理保留中止前已流式的文本；一次性与 continuable 的 `subagent/end` 边沿与 `SubagentResult.output` 一致。内容非空但不含文本的消息（例如仅含 reasoning）仍然优先于流式文本——规则针对的是内容为空，而非文本缺失。非空消息同样优先于**其后**才流式出的文本：子代理在流式后续步骤时被取消，报告的是更早那条完整消息，与 SDK 后端文档化的契约一致，截断由终止原因示意。三个包中的回归测试脚本化了空终止消息与取消路径，并在先前的选取实现下失败。
diff --git a/examples/acp-agent/tests/acp.snapshot.ts b/examples/acp-agent/tests/acp.snapshot.ts
index af044fc58f..852f9228ff 100644
--- a/examples/acp-agent/tests/acp.snapshot.ts
+++ b/examples/acp-agent/tests/acp.snapshot.ts
@@ -299,6 +299,13 @@ const SCENARIOS: Scenario[] = [
   // Windows bash process-tree kill is deferred with the Bash execution domain.
   { name: 'cancel-tool-calls', hasModelTurn: true, recorded: false, overridden: true, posixOnly: true },
   { name: 'subagent-spawn', hasModelTurn: true, recorded: true },
+  // Keyless, authored (like error-finish): a live child cannot be coaxed into
+  // a max-tokens step that assembled ONLY tool-call blocks — the truncation
+  // shape whose usage-only empty assistant/message must not erase the child's
+  // earlier text. The child fixture scripts text + todo_write, then a
+  // tool-only max-tokens cutoff; the parent's subagent tool result must carry
+  // the child's real partial answer with the max-tokens stop reason.
+  { name: 'subagent-max-tokens-partial', hasModelTurn: true, recorded: false },
   { name: 'subagent-multi', hasModelTurn: true, recorded: true },
   { name: 'subagent-fork', hasModelTurn: true, recorded: true },
   { name: 'subagent-mixed', hasModelTurn: true, recorded: true },
diff --git a/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/input.json b/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/input.json
new file mode 100644
index 0000000000..640bf92f7f
--- /dev/null
+++ b/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/input.json
@@ -0,0 +1,14 @@
+{
+  "steps": [
+    {
+      "op": "initialize"
+    },
+    {
+      "op": "newSession"
+    },
+    {
+      "op": "prompt",
+      "text": "Use the subagent tool exactly once to delegate this subtask: \"Write the words 'partial one', call todo_write once, then keep going until you are cut off.\" After the subagent returns, reply with the single word PARENT_DONE and stop."
+    }
+  ]
+}
diff --git a/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/session.1.jsonl b/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/session.1.jsonl
new file mode 100644
index 0000000000..55261e7a95
--- /dev/null
+++ b/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/session.1.jsonl
@@ -0,0 +1,30 @@
+{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":2,"cwd":"{{cwd}}","parentSession":"11111111-1111-4111-8111-111111111111","origin":"subagent","delegationDepth":1}
+{"type":"agent/inbox/spliced","seq":0,"time":1786348800126,"data":{"target":"next-turn","start":0,"inserted":[{"content":[{"type":"text","text":"Write the words 'partial one', call todo_write once, then keep going until you are cut off."}],"source":{"kind":"user"},"role":"user","id":"dbf0670a-79cc-4e2c-a298-c4d804e6fe61"}]}}
+{"type":"turn/start","seq":1,"time":1786348800126,"data":{"turn":1}}
+{"type":"agent/inbox/spliced","seq":2,"time":1786348800126,"data":{"target":"next-turn","start":0,"removedCount":1,"inserted":[]}}
+{"type":"subagent/descriptor","seq":3,"time":1786348800139,"data":{"version":2,"mode":"one-shot","provider":"spawn","label":"Truncated child"}}
+{"type":"step/start","seq":4,"time":1786348800142,"data":{"turn":1,"step":1}}
+{"type":"user/message","seq":5,"time":1786348800142,"data":{"content":[{"type":"text","text":"Write the words 'partial one', call todo_write once, then keep going until you are cut off."}],"source":{"kind":"user"},"role":"user","id":"dbf0670a-79cc-4e2c-a298-c4d804e6fe61"},"surfaceOp":"append"}
+{"type":"user/message","seq":6,"time":1786348800142,"data":{"content":[{"type":"text","text":"Current runtime context. This snapshot supersedes earlier runtime-context snapshots.\n\nCurrent DSH file policy: danger-full-access. The DSH file sandbox does not restrict file modifications by available operations.\n\nApproval prompts are disabled in this session: actions that require approval are rejected automatically — do not request sandbox escalation (do not set `sandbox_permissions`)."}],"source":{"kind":"plugin","plugin":"@deepseek-ai/dsh-system-prompt","form":"snapshot","sections":[{"name":"sandbox:policy","text":"Current DSH file policy: danger-full-access. The DSH file sandbox does not restrict file modifications by available operations."},{"name":"approval:policy","text":"Approval prompts are disabled in this session: actions that require approval are rejected automatically — do not request sandbox escalation (do not set `sandbox_permissions`)."}]},"role":"user","id":"ff5bdbe7-2eb0-4380-8edb-0e5c58ba9840"},"surfaceOp":"append"}
+{"type":"session/title","seq":7,"time":1786348800142,"data":{"title":"Write the words 'partial one',","messageSeqs":[5],"source":{"kind":"fallback"}}}
+{"type":"request/header","seq":8,"time":1786348800142,"data":{"header":{"config":{"provider":"deepseek-official","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}}
+{"type":"request/context","seq":9,"time":1786348800142,"data":{"provider":"deepseek-official","model":"deepseek-v4-flash"}}
+{"type":"assistant/chunk","seq":10,"time":0,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"text"}}}
+{"type":"assistant/chunk","seq":11,"time":1786348800146,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"partial one"}}}}
+{"type":"assistant/chunk","seq":12,"time":1786348800146,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":1,"blockType":"tool-call"}}}
+{"type":"assistant/chunk","seq":13,"time":1786348800146,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"tool-call","id":"call_child_1","name":"todo_write","arguments":"{\"todos\": [{\"content\": \"keep going\", \"status\": \"in_progress\"}]}"}}}}
+{"type":"assistant/chunk","seq":14,"time":1786348800146,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":20,"outputTokens":9}}}}
+{"type":"assistant/chunk","seq":15,"time":1786348800146,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
+{"type":"assistant/message","seq":16,"time":1786348800146,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"text","text":"partial one"},{"type":"tool-call","id":"call_child_1","name":"todo_write","arguments":"{\"todos\": [{\"content\": \"keep going\", \"status\": \"in_progress\"}]}"}],"source":{"kind":"model","provider":"deepseek-official","model":"deepseek-v4-flash"},"id":"5e4d07b2-6ce2-4ab6-8be0-fbdf2d3af138"},"usage":{"inputTokens":20,"outputTokens":9}},"sourceEventSeqs":[10,11,12,13,14,15],"surfaceOp":"append"}
+{"type":"tool/call","seq":17,"time":1786348800146,"data":{"turn":1,"step":1,"callId":"call_child_1","name":"todo_write","arguments":"{\"todos\": [{\"content\": \"keep going\", \"status\": \"in_progress\"}]}"}}
+{"type":"todo/write","seq":18,"time":1786348800150,"data":{"todos":[{"content":"keep going","status":"in_progress"}]}}
+{"type":"tool/result","seq":19,"time":1786348800151,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_child_1"},"content":[{"type":"tool-result","toolCallId":"call_child_1","content":[{"type":"text","text":"Updated todo list: 0 pending, 1 in progress, 0 completed."}],"isError":false}],"role":"user","id":"67efbbf3-ca1e-4d23-8f19-940cb391ff1e"}},"sourceEventSeqs":[17],"surfaceOp":"append"}
+{"type":"step/end","seq":20,"time":1786348800151,"data":{"turn":1,"step":1}}
+{"type":"step/start","seq":21,"time":1786348800156,"data":{"turn":1,"step":2}}
+{"type":"assistant/chunk","seq":22,"time":1786348800160,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"tool-call"}}}
+{"type":"assistant/chunk","seq":23,"time":1786348800160,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":0,"block":{"type":"tool-call","id":"call_child_2","name":"todo_write","arguments":"{\"todos\": [{\"content\": \"keep going\", \"status\": \"completed\"}]}"}}}}
+{"type":"assistant/chunk","seq":24,"time":1786348800160,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":30,"outputTokens":4}}}}
+{"type":"assistant/chunk","seq":25,"time":1786348800160,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"max-tokens"}}}}
+{"type":"assistant/message","seq":26,"time":1786348800160,"data":{"turn":1,"step":2,"message":{"role":"assistant","content":[],"source":{"kind":"model","provider":"deepseek-official","model":"deepseek-v4-flash"},"id":"bb92e4ec-f260-4415-9782-b71147ea378d"},"usage":{"inputTokens":30,"outputTokens":4}},"sourceEventSeqs":[22,23,24,25],"surfaceOp":"append"}
+{"type":"step/end","seq":27,"time":1786348800160,"data":{"turn":1,"step":2}}
+{"type":"turn/end","seq":28,"time":1786348800160,"data":{"turn":1,"reason":{"kind":"max-tokens"}}}
diff --git a/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/session.jsonl b/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/session.jsonl
new file mode 100644
index 0000000000..57386320e4
--- /dev/null
+++ b/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/session.jsonl
@@ -0,0 +1,26 @@
+{"type":"session","version":0,"id":"11111111-1111-4111-8111-111111111111","createdAt":1,"cwd":"{{cwd}}","delegationDepth":0}
+{"type":"agent/inbox/spliced","seq":0,"time":1786348800078,"data":{"target":"next-turn","start":0,"inserted":[{"content":[{"type":"text","text":"Use the subagent tool exactly once to delegate this subtask: \"Write the words 'partial one', call todo_write once, then keep going until you are cut off.\" After the subagent returns, reply with the single word PARENT_DONE and stop."}],"source":{"kind":"user"},"role":"user","id":"8787ce07-4f1f-4368-bf58-18e30484ed44"}]}}
+{"type":"turn/start","seq":1,"time":1786348800079,"data":{"turn":1}}
+{"type":"agent/inbox/spliced","seq":2,"time":1786348800079,"data":{"target":"next-turn","start":0,"removedCount":1,"inserted":[]}}
+{"type":"step/start","seq":3,"time":1786348800114,"data":{"turn":1,"step":1}}
+{"type":"user/message","seq":4,"time":1786348800114,"data":{"content":[{"type":"text","text":"Use the subagent tool exactly once to delegate this subtask: \"Write the words 'partial one', call todo_write once, then keep going until you are cut off.\" After the subagent returns, reply with the single word PARENT_DONE and stop."}],"source":{"kind":"user"},"role":"user","id":"8787ce07-4f1f-4368-bf58-18e30484ed44"},"surfaceOp":"append"}
+{"type":"user/message","seq":5,"time":1786348800114,"data":{"content":[{"type":"text","text":"Current runtime context. This snapshot supersedes earlier runtime-context snapshots.\n\nCurrent DSH file policy: danger-full-access. The DSH file sandbox does not restrict file modifications by available operations.\n\nApproval prompts are disabled in this session: actions that require approval are rejected automatically — do not request sandbox escalation (do not set `sandbox_permissions`)."}],"source":{"kind":"plugin","plugin":"@deepseek-ai/dsh-system-prompt","form":"snapshot","sections":[{"name":"sandbox:policy","text":"Current DSH file policy: danger-full-access. The DSH file sandbox does not restrict file modifications by available operations."},{"name":"approval:policy","text":"Approval prompts are disabled in this session: actions that require approval are rejected automatically — do not request sandbox escalation (do not set `sandbox_permissions`)."}]},"role":"user","id":"309b09a3-9593-4161-903d-cb5b14d8e9d9"},"surfaceOp":"append"}
+{"type":"session/title","seq":6,"time":1786348800114,"data":{"title":"Use the subagent tool exactly","messageSeqs":[4],"source":{"kind":"fallback"}}}
+{"type":"request/header","seq":7,"time":1786348800115,"data":{"header":{"config":{"provider":"deepseek-official","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}}
+{"type":"request/context","seq":8,"time":1786348800115,"data":{"provider":"deepseek-official","model":"deepseek-v4-flash"}}
+{"type":"assistant/chunk","seq":9,"time":1786348800120,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"tool-call"}}}
+{"type":"assistant/chunk","seq":10,"time":1786348800120,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"tool-call","id":"call_parent_1","name":"subagent","arguments":"{\"description\": \"Truncated child\", \"prompt\": \"Write the words 'partial one', call todo_write once, then keep going until you are cut off.\"}"}}}}
+{"type":"assistant/chunk","seq":11,"time":1786348800120,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":10,"outputTokens":5}}}}
+{"type":"assistant/chunk","seq":12,"time":1786348800120,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
+{"type":"assistant/message","seq":13,"time":1786348800120,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"tool-call","id":"call_parent_1","name":"subagent","arguments":"{\"description\": \"Truncated child\", \"prompt\": \"Write the words 'partial one', call todo_write once, then keep going until you are cut off.\"}"}],"source":{"kind":"model","provider":"deepseek-official","model":"deepseek-v4-flash"},"id":"f4269cd2-9132-4b68-8f9b-ff3a40321bc9"},"usage":{"inputTokens":10,"outputTokens":5}},"sourceEventSeqs":[9,10,11,12],"surfaceOp":"append"}
+{"type":"tool/call","seq":14,"time":1786348800121,"data":{"turn":1,"step":1,"callId":"call_parent_1","name":"subagent","arguments":"{\"description\": \"Truncated child\", \"prompt\": \"Write the words 'partial one', call todo_write once, then keep going until you are cut off.\"}"}}
+{"type":"tool/result","seq":15,"time":1786348800163,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_parent_1"},"content":[{"type":"tool-result","toolCallId":"call_parent_1","content":[{"type":"text","text":"Error: subagent run hit its token limit before finishing\nPartial output before the run ended:\npartial one"}],"isError":true}],"role":"user","id":"5dd34050-a533-4f1b-99ee-5fc62c6a4502"}},"sourceEventSeqs":[14],"surfaceOp":"append"}
+{"type":"step/end","seq":16,"time":1786348800163,"data":{"turn":1,"step":1}}
+{"type":"step/start","seq":17,"time":1786348800169,"data":{"turn":1,"step":2}}
+{"type":"assistant/chunk","seq":18,"time":1786348800173,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"text"}}}
+{"type":"assistant/chunk","seq":19,"time":1786348800173,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"PARENT_DONE"}}}}
+{"type":"assistant/chunk","seq":20,"time":1786348800173,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":12,"outputTokens":2}}}}
+{"type":"assistant/chunk","seq":21,"time":1786348800173,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
+{"type":"assistant/message","seq":22,"time":1786348800173,"data":{"turn":1,"step":2,"message":{"role":"assistant","content":[{"type":"text","text":"PARENT_DONE"}],"source":{"kind":"model","provider":"deepseek-official","model":"deepseek-v4-flash"},"id":"fb14560d-1d98-4b18-8736-b079de400315"},"usage":{"inputTokens":12,"outputTokens":2}},"sourceEventSeqs":[18,19,20,21],"surfaceOp":"append"}
+{"type":"step/end","seq":23,"time":1786348800173,"data":{"turn":1,"step":2}}
+{"type":"turn/end","seq":24,"time":1786348800173,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/stdout.expected.jsonl b/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/stdout.expected.jsonl
new file mode 100644
index 0000000000..a460e019d4
--- /dev/null
+++ b/examples/acp-agent/tests/snapshots/subagent-max-tokens-partial/stdout.expected.jsonl
@@ -0,0 +1,4 @@
+{"jsonrpc":"2.0","id":1,"result":{"protocolVersion":1,"agentInfo":{"name":"deepseek-harness-acp","version":"0.0.1"},"agentCapabilities":{"promptCapabilities":{"image":false,"audio":false,"embeddedContext":false}},"authMethods":[]}}
+{"jsonrpc":"2.0","id":2,"result":{"sessionId":"{{sessionId}}"}}
+{"jsonrpc":"2.0","method":"session/update","params":{"sessionId":"{{sessionId}}","update":{"sessionUpdate":"agent_message_chunk","content":{"type":"text","text":"PARENT_DONE"}}}}
+{"jsonrpc":"2.0","id":3,"result":{"stopReason":"end_turn"}}
diff --git a/packages/subagent/subagent-acp/src/run.ts b/packages/subagent/subagent-acp/src/run.ts
index 17476be41c..c57643411a 100644
--- a/packages/subagent/subagent-acp/src/run.ts
+++ b/packages/subagent/subagent-acp/src/run.ts
@@ -24,6 +24,7 @@ import {
 } from '@agentclientprotocol/sdk'
 import type { ContentBlock } from '@deepseek-ai/dsh-llm'
 import { SessionId } from '@deepseek-ai/dsh-session'
+import { AssistantOutputFold } from '@deepseek-ai/dsh-subagent'
 import type { SubagentResult, SubagentRun, SubagentStartRequest, SubagentStopReason } from '@deepseek-ai/dsh-subagent'
 import type { SubprocessHandle, SubprocessSpawnSpec } from '@deepseek-ai/dsh-subprocess'
 
@@ -232,8 +233,10 @@ export async function startAcpRun(request: SubagentStartRequest, spec: AcpRunSpe
   let processDisposal: Promise<void> | undefined
   const disposeProcess = (): Promise<void> => (processDisposal ??= disposeAcpChild(child, spec.disposeEofGraceMs))
 
-  // Accumulate the child's streamed assistant text — the SubagentResult output.
-  const output: string[] = []
+  // The child's streamed assistant text, accumulated under the seam's
+  // canonical selection rule (`AssistantOutputFold`); ACP surfaces no complete
+  // assistant messages, so only the streamed-fallback half applies.
+  const fold = new AssistantOutputFold()
   // Shared mutable state keeps cancellation visible across async closures.
   const flags = { cancelled: false }
 
@@ -241,7 +244,7 @@ export async function startAcpRun(request: SubagentStartRequest, spec: AcpRunSpe
     sessionUpdate(params: SessionNotification): Promise<void> {
       const update = params.update
       if (update.sessionUpdate === 'agent_message_chunk') {
-        output.push(acpContentText(update.content))
+        fold.pushText(acpContentText(update.content))
       }
       // Other updates (thoughts, tool calls, plans) are consumed but not
       // surfaced — the subagent returns only its final answer.
@@ -284,13 +287,8 @@ export async function startAcpRun(request: SubagentStartRequest, spec: AcpRunSpe
   const onAbort = (): void => { requestCancel() }
   request.signal.addEventListener('abort', onAbort, { once: true })
 
-  // The accumulated child text as harness ContentBlocks (empty array when the
-  // child streamed nothing). Read at every return so a partial answer survives
-  // a later cancel/error.
-  const collectOutput = (): ContentBlock[] => {
-    const text = output.join('')
-    return text.length > 0 ? [{ type: 'text', text }] : []
-  }
+  // Read at every return so a partial answer survives a later cancel/error.
+  const collectOutput = (): ContentBlock[] => fold.collect() ?? []
 
   // Establish the remote session before publishing a handle. Any failure owns
   // the still-private process and therefore reaps it before rejecting.
diff --git a/packages/subagent/subagent-dsh-sdk/src/run.ts b/packages/subagent/subagent-dsh-sdk/src/run.ts
index 39bf53fffb..194ce3badf 100644
--- a/packages/subagent/subagent-dsh-sdk/src/run.ts
+++ b/packages/subagent/subagent-dsh-sdk/src/run.ts
@@ -16,7 +16,7 @@ import { DeepSeekHarness, type HarnessNotification } from '@deepseek-ai/dsh-sdk-
 import type { ContentBlock } from '@deepseek-ai/dsh-llm'
 import { SessionId, type SessionEvent, type TurnEndReason } from '@deepseek-ai/dsh-session'
 import type { SubagentResult, SubagentRun, SubagentStartRequest, SubagentStopReason } from '@deepseek-ai/dsh-subagent'
-import { assistantMessageOutput, settleRunResult, subprocessRunHandle } from '@deepseek-ai/dsh-subagent'
+import { AssistantOutputFold, settleRunResult, subprocessRunHandle } from '@deepseek-ai/dsh-subagent'
 import { scrubbedParentEnv } from '@deepseek-ai/dsh-subprocess'
 
 /** Resolved spawn spec for an SDK runtime child process (no defaults — see Config). */
@@ -163,28 +163,14 @@ export async function startSdkRun(request: SubagentStartRequest, spec: SdkRunSpe
   }
 
   const childSessionId = `session-${randomUUID().replaceAll('-', '')}`
-  // The child's final answer, folded incrementally under the seam's canonical
-  // rule (`finalAssistantOutput`): the last NON-EMPTY complete assistant
-  // message when one exists, else the text streamed so far (a partial answer
-  // surviving cancel). An empty-content message hosts only usage (a max-tokens
-  // step that assembled no text blocks), so it never erases streamed text.
-  let lastMessage: ContentBlock[] | undefined
-  const partial: string[] = []
+  // The child's final answer under the seam's canonical selection rule
+  // (`AssistantOutputFold`); a partial answer survives cancel and error paths.
+  const fold = new AssistantOutputFold()
   const observe = (notification: HarnessNotification): void => {
     if (notification.method !== 'session.event' || notification.params.sessionId !== childSessionId) return
-    const event = notification.params.event as SessionEvent
-    const content = assistantMessageOutput(event)
-    if (content !== undefined) {
-      lastMessage = content
-    } else if (event.type === 'assistant/chunk' && event.data.chunk.type === 'text-delta') {
-      partial.push(event.data.chunk.text)
-    }
-  }
-  const collectOutput = (): ContentBlock[] => {
-    if (lastMessage !== undefined) return lastMessage
-    const text = partial.join('')
-    return text.length > 0 ? [{ type: 'text', text }] : []
+    fold.push(notification.params.event as SessionEvent)
   }
+  const collectOutput = (): ContentBlock[] => fold.collect() ?? []
 
   // Race the child turn against local cancellation; the shared settlement
   // flattens failures under the seam's never-reject contract.
diff --git a/packages/subagent/subagent-inprocess/src/index.ts b/packages/subagent/subagent-inprocess/src/index.ts
index d621674a6e..0f09e8cfa4 100644
--- a/packages/subagent/subagent-inprocess/src/index.ts
+++ b/packages/subagent/subagent-inprocess/src/index.ts
@@ -220,9 +220,7 @@ function readResult(
 ): SubagentResult {
   const own = child.session.events.slice(boundary)
   const lastEnd = findLastMessageTurnEnd(own)
-  // Canonical selection (`finalAssistantOutput`): the last non-empty assistant
-  // message, else the text streamed before cancel/error/truncation cut the
-  // turn short — an empty usage-only message never erases real output.
+  // The seam's canonical selection rule; a partial answer survives cancel and truncation.
   const output: ContentBlock[] = finalAssistantOutput(own) ?? []
   const recorded = toStopReason(lastEnd?.data.reason)
   // Disposal can tear the owner down before the loop records its ordinary
diff --git a/packages/subagent/subagent/README.i18n.yaml b/packages/subagent/subagent/README.i18n.yaml
index 2b47a91a4b..81fcd7d10b 100644
--- a/packages/subagent/subagent/README.i18n.yaml
+++ b/packages/subagent/subagent/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/subagent/subagent/README.md
-README.md: d2d5356fd82a47ecf5cd6b633e5338dde7047901
-README.zh.md: 2fdc3ae6e8376ef7c7aaf11c8e85dd909ef92d2c
+README.md: 843fd0af4a86ea3a10d4a4ee101aa05478a2300e
+README.zh.md: 497f2a928c8eaeff8ef0486f5344c3d257be7391
diff --git a/packages/subagent/subagent/README.md b/packages/subagent/subagent/README.md
index d2d5356fd8..843fd0af4a 100644
--- a/packages/subagent/subagent/README.md
+++ b/packages/subagent/subagent/README.md
@@ -56,7 +56,7 @@ The seam owns the depth vocabulary shared by Service providers and Consumers: th
 
 `provider.start(request): Promise<SubagentRun>` is the ownership-transfer boundary; the delegation tool also uses it inside its one-shot Task-backed background path. Before fulfillment, the provider owns setup and must cancel, roll back, and quiesce unpublished resources on every failure. After fulfillment, the caller owns the run and must call `dispose()` on every path; remaining prompt and turn work belongs to `SubagentRun.result`.
 
-`SubagentRun.result` resolves to `{ output, structured?, stopReason }`. Child-level failures resolve with a non-`completed` reason; only an infrastructure fault that the seam cannot represent may reject. `dispose()` is idempotent, cancels remaining work, and waits for both result settlement and child-resource quiescence. A result rejection remains on `result`; `dispose()` rejects only for an independent resource-release failure. `output` and the `subagent/end` edge's `lastAssistantMessage` share one selection rule, implemented by the exported `finalAssistantOutput` helper: the child's last non-empty assistant message, else the text it streamed before the turn was cut short ([`SubagentResult.output`](../../../docs/subsystems/subagent.md#the-terminal-result-subagentresult) owns the contract).
+`SubagentRun.result` resolves to `{ output, structured?, stopReason }`. Child-level failures resolve with a non-`completed` reason; only an infrastructure fault that the seam cannot represent may reject. `dispose()` is idempotent, cancels remaining work, and waits for both result settlement and child-resource quiescence. A result rejection remains on `result`; `dispose()` rejects only for an independent resource-release failure. `output` and the `subagent/end` edge's `lastAssistantMessage` share one selection rule, implemented once by the exported `AssistantOutputFold`/`finalAssistantOutput` helpers: the child's last non-empty assistant message, else the text it streamed before the turn was cut short ([`SubagentResult.output`](../../../docs/subsystems/subagent.md#the-terminal-result-subagentresult) owns the contract).
 
 A local run publishes an ordinary child agent/session before `start()` fulfills, returns that shared session id as `SubagentRun.id`, exposes the exact child as `SubagentRun.localAgent`, records `request.parent.session.id` in the child's `parentSession` header, and appends the resolved descriptor inside its initial turn. Remote providers instead mint a parent-scoped lifecycle id and return `localAgent: undefined`; without a local child session, their one-shot runs are not part of trace-backed enumeration.
 
diff --git a/packages/subagent/subagent/README.zh.md b/packages/subagent/subagent/README.zh.md
index 2fdc3ae6e8..497f2a928c 100644
--- a/packages/subagent/subagent/README.zh.md
+++ b/packages/subagent/subagent/README.zh.md
@@ -56,7 +56,7 @@ subagent seam 允许一个 agent（智能体）通过具名提供方把工作委
 
 `provider.start(request): Promise<SubagentRun>` 是所有权转移边界；委派工具也会在其由 Task 支撑的一次性后台路径中使用它。兑现前，提供方拥有设置过程，并且每次失败时都必须取消、回滚并使未发布资源完全停稳。兑现后，调用方拥有该运行，并且必须在每条路径上调用 `dispose()`；剩余提示词和轮次工作属于 `SubagentRun.result`。
 
-`SubagentRun.result` 兑现为 `{ output, structured?, stopReason }`。子 agent 级失败会以非 `completed` 原因兑现；只有 seam 无法表示的基础设施故障才可以拒绝。`dispose()` 是幂等的，会取消剩余工作，并等待结果结算以及子 agent 资源完全停稳。`result` 的 rejection 仍归 `result` 通道；只有独立的资源释放失败会使 `dispose()` 拒绝。`output` 与 `subagent/end` 边沿的 `lastAssistantMessage` 共用同一条选取规则，由导出的 `finalAssistantOutput` 辅助函数实现：取子 agent 最后一条非空 assistant 消息，否则取轮次被截断前已流式的文本（契约归 [`SubagentResult.output`](../../../docs/subsystems/subagent.md#the-terminal-result-subagentresult) 所有）。
+`SubagentRun.result` 兑现为 `{ output, structured?, stopReason }`。子 agent 级失败会以非 `completed` 原因兑现；只有 seam 无法表示的基础设施故障才可以拒绝。`dispose()` 是幂等的，会取消剩余工作，并等待结果结算以及子 agent 资源完全停稳。`result` 的 rejection 仍归 `result` 通道；只有独立的资源释放失败会使 `dispose()` 拒绝。`output` 与 `subagent/end` 边沿的 `lastAssistantMessage` 共用同一条选取规则，由导出的 `AssistantOutputFold`／`finalAssistantOutput` 辅助函数唯一实现：取子 agent 最后一条非空 assistant 消息，否则取轮次被截断前已流式的文本（契约归 [`SubagentResult.output`](../../../docs/subsystems/subagent.md#the-terminal-result-subagentresult) 所有）。
 
 本地运行会在 `start()` 兑现前发布普通的子 agent／会话，把该共享会话 id 作为 `SubagentRun.id` 返回，以 `SubagentRun.localAgent` 公开准确的子 agent，把 `request.parent.session.id` 记录到子 agent 的 `parentSession` header，并在其初始轮次内追加已解析的描述符。远程提供方则生成 parent 作用域的生命周期 id，并返回 `localAgent: undefined`；由于没有本地 child 会话，其一次性运行不会进入基于追踪的枚举结果。
 
diff --git a/packages/subagent/subagent/src/assistant-output.ts b/packages/subagent/subagent/src/assistant-output.ts
index 5fea60fa89..5b11031c19 100644
--- a/packages/subagent/subagent/src/assistant-output.ts
+++ b/packages/subagent/subagent/src/assistant-output.ts
@@ -1,12 +1,12 @@
 /**
- * Canonical selection of a child's final assistant output from its session
- * events. Every surface that reports "the child's answer" — backend run
- * results and `subagent/end.lastAssistantMessage` — applies this one rule so
- * observers agree: the last NON-EMPTY assistant message wins; an empty-content
- * message hosts only usage (the loop appends one when a max-tokens step
- * assembled no executable blocks) and never erases real output; without any
- * non-empty message, the text streamed so far is the answer (a partial
- * surviving cancel, error, and truncation paths).
+ * Canonical selection of a child's final assistant output. Every surface that
+ * reports "the child's answer" — backend run results and
+ * `subagent/end.lastAssistantMessage` — applies this one rule so observers
+ * agree: the last NON-EMPTY assistant message wins; an empty-content message
+ * hosts only usage (the loop appends one when a max-tokens step assembled no
+ * executable blocks) and never erases real output; without any non-empty
+ * message, the text streamed so far is the answer (a partial surviving
+ * cancel, error, and truncation paths).
  *
  * @module @deepseek-ai/dsh-subagent/assistant-output
  */
@@ -15,37 +15,57 @@ import type { ContentBlock } from '@deepseek-ai/dsh-llm'
 import type { SessionEvent } from '@deepseek-ai/dsh-session'
 
 /**
- * The content one event contributes as a candidate final answer: an
- * `assistant/message` with non-empty content. An empty-content message hosts
- * only usage and contributes none.
- * @param event - any session event.
- * @returns the message content, or `undefined` when this event is not a
- *   non-empty assistant message.
+ * Incremental fold of the selection rule, for backends that observe a child's
+ * output as it streams: session-event backends {@link push} each event, and
+ * transports without session events (ACP content chunks) {@link pushText} raw
+ * text into the same streamed fallback.
  */
-export function assistantMessageOutput(event: SessionEvent): ContentBlock[] | undefined {
-  if (event.type !== 'assistant/message') return undefined
-  const content = event.data.message.content
-  return content.length > 0 ? content : undefined
+export class AssistantOutputFold {
+  private message: ContentBlock[] | undefined
+  private partial: string[] = []
+
+  /**
+   * Fold one session event: a non-empty assistant message becomes the
+   * candidate final answer, and a `text-delta` chunk extends the streamed
+   * fallback; every other event contributes nothing.
+   * @param event - the next observed session event.
+   */
+  push(event: SessionEvent): void {
+    if (event.type === 'assistant/message') {
+      const content = event.data.message.content
+      if (content.length > 0) this.message = content
+    } else if (event.type === 'assistant/chunk' && event.data.chunk.type === 'text-delta') {
+      this.partial.push(event.data.chunk.text)
+    }
+  }
+
+  /**
+   * Extend the streamed fallback with text observed outside session events.
+   * @param text - the next streamed text piece (an empty piece is a no-op).
+   */
+  pushText(text: string): void {
+    this.partial.push(text)
+  }
+
+  /**
+   * Select the final output folded so far.
+   * @returns the last non-empty assistant message, else the accumulated
+   *   streamed text, or `undefined` when the child produced neither.
+   */
+  collect(): ContentBlock[] | undefined {
+    if (this.message !== undefined) return this.message
+    const text = this.partial.join('')
+    return text.length > 0 ? [{ type: 'text', text }] : undefined
+  }
 }
 
 /**
- * Select the final assistant output from one child-owned event suffix: the
- * last non-empty assistant message, else the accumulated `text-delta` stream.
+ * Apply the selection rule to one complete child-owned event suffix.
  * @param events - the child-owned events (after any seed or epoch boundary).
  * @returns the selected output, or `undefined` when the child produced none.
  */
 export function finalAssistantOutput(events: readonly SessionEvent[]): ContentBlock[] | undefined {
-  let message: ContentBlock[] | undefined
-  const partial: string[] = []
-  for (const event of events) {
-    const content = assistantMessageOutput(event)
-    if (content !== undefined) {
-      message = content
-    } else if (event.type === 'assistant/chunk' && event.data.chunk.type === 'text-delta') {
-      partial.push(event.data.chunk.text)
-    }
-  }
-  if (message !== undefined) return message
-  const text = partial.join('')
-  return text.length > 0 ? [{ type: 'text', text }] : undefined
+  const fold = new AssistantOutputFold()
+  for (const event of events) fold.push(event)
+  return fold.collect()
 }
diff --git a/packages/subagent/subagent/src/index.ts b/packages/subagent/subagent/src/index.ts
index d3e26e6dbc..e39d5eec70 100644
--- a/packages/subagent/subagent/src/index.ts
+++ b/packages/subagent/subagent/src/index.ts
@@ -69,7 +69,7 @@ import { snapshotSubagentDescriptor } from './descriptor.ts'
 import { subagentIdentityProjectionDefinition, subagentTimingProjectionDefinition } from './projection.ts'
 
 export * from './out-of-process.ts'
-export { assistantMessageOutput, finalAssistantOutput } from './assistant-output.ts'
+export { AssistantOutputFold, finalAssistantOutput } from './assistant-output.ts'
 export { SubagentRunId } from './types.ts'
 export type {
   ContinuableCreateRequest,
diff --git a/packages/subagent/subagent/src/lifecycle.ts b/packages/subagent/subagent/src/lifecycle.ts
index 26fdd54256..df050f5538 100644
--- a/packages/subagent/subagent/src/lifecycle.ts
+++ b/packages/subagent/subagent/src/lifecycle.ts
@@ -129,7 +129,9 @@ export function observeRun(
       emit('subagent/end', {
         ...identity,
         stopReason: result.stopReason,
-        lastAssistantMessage: result.output,
+        // One encoding for "no output" across both lifecycle shapes: the
+        // field is absent, matching the continuable epoch edge.
+        ...result.output.length === 0 ? {} : { lastAssistantMessage: result.output },
       }, parent)
     },
     () => {
diff --git a/packages/subagent/subagent/tests/assistant-output.spec.ts b/packages/subagent/subagent/tests/assistant-output.spec.ts
index 5219209249..2205431aae 100644
--- a/packages/subagent/subagent/tests/assistant-output.spec.ts
+++ b/packages/subagent/subagent/tests/assistant-output.spec.ts
@@ -1,7 +1,7 @@
 import { describe, expect, it } from 'vitest'
 import type { ContentBlock } from '@deepseek-ai/dsh-llm'
 import type { SessionEvent } from '@deepseek-ai/dsh-session'
-import { assistantMessageOutput, finalAssistantOutput } from '../src/assistant-output.ts'
+import { AssistantOutputFold, finalAssistantOutput } from '../src/assistant-output.ts'
 
 function message(content: ContentBlock[]): SessionEvent {
   return { type: 'assistant/message', data: { message: { content } } } as SessionEvent
@@ -15,15 +15,6 @@ function reasoningDelta(text: string): SessionEvent {
   return { type: 'assistant/chunk', data: { chunk: { type: 'reasoning-delta', text } } } as SessionEvent
 }
 
-describe('assistantMessageOutput', () => {
-  it('returns content only for a non-empty assistant message', () => {
-    const content: ContentBlock[] = [{ type: 'text', text: 'answer' }]
-    expect(assistantMessageOutput(message(content))).toBe(content)
-    expect(assistantMessageOutput(message([]))).toBeUndefined()
-    expect(assistantMessageOutput(textDelta('chunk'))).toBeUndefined()
-  })
-})
-
 describe('finalAssistantOutput', () => {
   it('selects the last non-empty message past a later empty usage-only message', () => {
     const events = [
@@ -58,3 +49,17 @@ describe('finalAssistantOutput', () => {
     expect(finalAssistantOutput([reasoningDelta('thinking'), message([])])).toBeUndefined()
   })
 })
+
+describe('AssistantOutputFold', () => {
+  it('folds raw text pieces into the same streamed fallback (ACP chunk transport)', () => {
+    const fold = new AssistantOutputFold()
+    fold.pushText('partial ')
+    fold.pushText('')
+    fold.pushText('answer')
+    expect(fold.collect()).toEqual([{ type: 'text', text: 'partial answer' }])
+  })
+
+  it('collects undefined until any output is folded', () => {
+    expect(new AssistantOutputFold().collect()).toBeUndefined()
+  })
+})
diff --git a/packages/subagent/subagent/tests/service.spec.ts b/packages/subagent/subagent/tests/service.spec.ts
index a50696cf2a..b46f2489fa 100644
--- a/packages/subagent/subagent/tests/service.spec.ts
+++ b/packages/subagent/subagent/tests/service.spec.ts
@@ -15,6 +15,7 @@ import SubagentService, {
   type SubagentProvider,
   type SubagentResult,
   type SubagentRun,
+  type SubagentRunEndInfo,
   type SubagentStartRequest,
 } from '@deepseek-ai/dsh-subagent'
 import { SessionId, type SessionEvent } from '@deepseek-ai/dsh-session'
@@ -263,6 +264,17 @@ describe('SubagentService', () => {
       stopReason: 'completed',
     }))
 
+    // "No output" has ONE encoding on the end edge: the field is absent,
+    // never an empty array, matching the continuable epoch edge.
+    const silent = new StubProvider('silent', NO_CAPS, { output: [], stopReason: 'completed' })
+    subagents.registerProvider(silent)
+    const silentRun = await subagents.start('silent', baseRequest())
+    await silentRun.result
+    await Promise.resolve()
+    const silentEnd = ended.mock.calls.map(call => call[0] as SubagentRunEndInfo).find(info => info.provider === 'silent')
+    expect(silentEnd).toBeDefined()
+    expect('lastAssistantMessage' in silentEnd!).toBe(false)
+
     const failure = Promise.withResolvers<SubagentResult>()
     subagents.registerProvider({
       name: 'infra',
diff --git a/packages/subagent/tool-subagent/README.i18n.yaml b/packages/subagent/tool-subagent/README.i18n.yaml
index e5f0c43f61..8c41c6413a 100644
--- a/packages/subagent/tool-subagent/README.i18n.yaml
+++ b/packages/subagent/tool-subagent/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/subagent/tool-subagent/README.md
-README.md: 6ec313b3b97f0ffa7488025d4314b1c6231a6f6a
-README.zh.md: 1fd88363b3ade9d57c194580f81295eed139ac50
+README.md: ac3ec0563cce9128608ca31860b034a103dc1a3a
+README.zh.md: d64831d7cf64800ad3307ce6cb7f294500a0a6f0
diff --git a/packages/subagent/tool-subagent/README.md b/packages/subagent/tool-subagent/README.md
index 6ec313b3b9..ac3ec0563c 100644
--- a/packages/subagent/tool-subagent/README.md
+++ b/packages/subagent/tool-subagent/README.md
@@ -8,7 +8,7 @@ The model-facing delegation tool over one configured `ctx.subagents` provider. C
 
 Each plugin instance binds one `provider` to one `toolName`; the model receives no provider selector. Load another distinctly named instance to expose another transport. The tool registers only while its provider exists, avoiding sibling load-order and provider-reload dependencies. Its description follows `provider.inheritsParentContext`: fresh children require standalone prompts, while forked children already see completed parent turns.
 
-A foreground call passes the execution signal through startup and execution, awaits `run.result`, and always awaits `run.dispose()` before returning. Only `completed` returns the canonical `{ kind: 'foreground', runId, output: JsonValue[] }`, rendered as the same final text; abort, refusal, token limit, and other failures become errored tool results without partial output. If result collection and disposal both reject, the errored result preserves both diagnostics.
+A foreground call passes the execution signal through startup and execution, awaits `run.result`, and always awaits `run.dispose()` before returning. Only `completed` returns the canonical `{ kind: 'foreground', runId, output: JsonValue[] }`, rendered as the same final text; abort, refusal, token limit, and other failures become errored tool results whose message appends the child's preserved partial text (the `SubagentResult.output` selection) after the stop-reason headline, so a truncated answer is never reported as success yet never silently lost. If result collection and disposal both reject, the errored result preserves both diagnostics.
 
 With `run_in_background: true`, `backgroundMode` selects the route. `one-shot` registers a plain parent-owned Task and returns canonical `{ kind: 'background', taskId }`, rendered as `started background subagent task <id>`, even when the provider supports continuable children; generic task tools own its later status, collection, cancellation, and notices. `continuable` requires a provider with the `prepareContinuable` capability, calls `ctx.subagents.startContinuable()`, and returns `{ kind: 'continuable', subagentId }`, rendered as `started subagent <childId>`. The continuable route resolves at inbox acceptance: the child owns its own turns from there, so this call neither waits for nor collects a result, and the child does not report back — its transcript by that id is the source of its output, and the optional global `send_message` tool sends it more work. Starting continuable work does not require `send_message` to be loaded. See the [background subagent Agent Note](../../../.agents/notes/implemented/feature/2026-07-08-background-subagent-tasks.md), the [continuable subagents Agent Note](../../../.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.md), and the [merged-service Agent Note](../../../.agents/notes/implemented/simplification/2026-07-26-merge-subagent-control-service.md).
 
diff --git a/packages/subagent/tool-subagent/README.zh.md b/packages/subagent/tool-subagent/README.zh.md
index 1fd88363b3..d64831d7cf 100644
--- a/packages/subagent/tool-subagent/README.zh.md
+++ b/packages/subagent/tool-subagent/README.zh.md
@@ -8,7 +8,7 @@
 
 每个插件实例把一个 `provider` 绑定到一个 `toolName`；模型不会收到提供方选择器。如需公开另一种传输，请加载另一个名称不同的实例。工具只在其提供方存在时注册，从而避免对同级加载顺序和提供方重新加载的依赖。工具描述遵循 `provider.inheritsParentContext`：新建子 agent（智能体）需要独立提示词，而 fork 子 agent 已能看到父级已完成轮次。
 
-前台调用会让执行信号贯穿启动和执行，等待 `run.result`，并且在返回前总会等待 `run.dispose()`。只有 `completed` 会返回规范值 `{ kind: 'foreground', runId, output: JsonValue[] }`，并渲染为相同的最终文本；中止、拒绝、token 上限和其他失败都会变成出错的工具结果，不包含局部输出。如果结果收集与 dispose（资源释放）都 reject，出错的结果会保留两项诊断信息。
+前台调用会让执行信号贯穿启动和执行，等待 `run.result`，并且在返回前总会等待 `run.dispose()`。只有 `completed` 会返回规范值 `{ kind: 'foreground', runId, output: JsonValue[] }`，并渲染为相同的最终文本；中止、拒绝、token 上限和其他失败都会变成出错的工具结果，其消息在终止原因标题之后附带子代理保留下来的部分文本（即 `SubagentResult.output` 的选取结果）——被截断的回答不会被报告为成功，也绝不会被悄悄丢弃。如果结果收集与 dispose（资源释放）都 reject，出错的结果会保留两项诊断信息。
 
 设置 `run_in_background: true` 后，`backgroundMode` 会选择路由。`one-shot` 会注册一个归父级所有的普通 Task，并返回规范值 `{ kind: 'background', taskId }`，渲染为 `started background subagent task <id>`，即使提供方支持可继续子 agent 也不例外；通用 Task 工具负责其后续状态、收集、取消和通知。`continuable` 要求提供方具备 `prepareContinuable` 能力，调用 `ctx.subagents.startContinuable()`，并返回 `{ kind: 'continuable', subagentId }`，渲染为 `started subagent <childId>`。可继续路由在 inbox 接受时结算：子 agent 自此拥有自己的轮次，因此该调用既不等待也不收集结果，而且子 agent 不会回报——通过该 id 查看其 transcript（文本记录）即是其输出来源，可选的全局 `send_message` 工具则向其发送更多工作。启动可继续工作不要求加载 `send_message`。见 [后台 subagent Agent Note](../../../.agents/notes/implemented/feature/2026-07-08-background-subagent-tasks.md)、[可继续的 subagent Agent Note](../../../.agents/notes/implemented/feature/2026-07-28-continuable-subagent-conversations.md)和[服务合并 Agent Note](../../../.agents/notes/implemented/simplification/2026-07-26-merge-subagent-control-service.md)。
 
diff --git a/packages/subagent/tool-subagent/src/index.ts b/packages/subagent/tool-subagent/src/index.ts
index 67894c32cb..b2743054c4 100644
--- a/packages/subagent/tool-subagent/src/index.ts
+++ b/packages/subagent/tool-subagent/src/index.ts
@@ -134,6 +134,21 @@ function stopReasonError(result: SubagentResult): string | undefined {
   }
 }
 
+/**
+ * Append the child's preserved partial answer to a stop-reason error so a
+ * truncated or cancelled child's real text still reaches the parent model.
+ * @param error - the stop-reason headline.
+ * @param output - the child's selected output (`SubagentResult.output`).
+ * @returns the headline, extended with the partial text when any exists.
+ */
+function withPartialText(error: string, output: ContentBlock[]): string {
+  const text = output
+    .filter((block): block is Extract<ContentBlock, { type: 'text' }> => block.type === 'text')
+    .map(block => block.text)
+    .join('')
+  return text.length === 0 ? error : `${error}\nPartial output before the run ended:\n${text}`
+}
+
 type ForegroundToolResult = {
   readonly kind: 'foreground'
   readonly runId: SubagentRun['id']
@@ -149,8 +164,9 @@ async function settleForegroundRun(run: SubagentRun): Promise<ForegroundToolResu
     run.result.then((result): ForegroundToolResult => {
       const error = stopReasonError(result)
       if (error !== undefined) {
-        // The registry converts this throw to isError; partial output is not success.
-        throw new Error(error)
+        // The registry converts this throw to isError; partial output is not
+        // success, but the preserved partial answer still reaches the parent.
+        throw new Error(withPartialText(error, result.output))
       }
       return {
         kind: 'foreground',
diff --git a/packages/subagent/tool-subagent/tests/tool-subagent.spec.ts b/packages/subagent/tool-subagent/tests/tool-subagent.spec.ts
index 91dc423cd4..ed83696358 100644
--- a/packages/subagent/tool-subagent/tests/tool-subagent.spec.ts
+++ b/packages/subagent/tool-subagent/tests/tool-subagent.spec.ts
@@ -154,6 +154,9 @@ describe('dsh-tool-subagent', () => {
     const result = await callSubagent(ctx, { description: 'd', prompt: 'p' })
     expect(result.isError).toBe(true)
     expect(text(result)).toContain(fragment)
+    // The failure is not partial success, but the child's preserved partial
+    // answer still reaches the parent model inside the error result.
+    expect(text(result)).toContain('scripted subagent reply')
   })
 
   it('registers under a configurable toolName so multiple providers can coexist', async () => {