@deepseek-ai/dsh-goal-round-driver 0.2.1-rc.5 → 0.2.2-alpha.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.i18n.yaml CHANGED
@@ -24,8 +24,8 @@
24
24
  en: a1312339977f1127
25
25
  zh: 57450327ff7e7d83
26
26
  /deepseek-ai-dsh-goal-round-driver/use-this-package/when-continuation-stops:
27
- en: 32ad77251e9f5acd
28
- zh: a72f4c3e95c1a137
27
+ en: f3cb23a11176dc4f
28
+ zh: de7aa54d5c93445d
29
29
  /deepseek-ai-dsh-goal-round-driver/use-this-package/after-resume-fork-or-unload:
30
30
  en: 85d5470b1b330330
31
31
  zh: e761f9046ba7af8a
package/README.md CHANGED
@@ -50,7 +50,7 @@ With an exact live agent idle, an active armed goal, and remaining capacity, the
50
50
 
51
51
  ### When continuation stops
52
52
 
53
- A round starts only at whole-agent idle and, when the subagent service is loaded, only after the agent's progressing descendants settle; completion, pause, and blocking suppress continuation; a host-initiated pause also aborts the turn already running, while a model-initiated pause inside its own turn finishes normally. An edit only invalidates an in-flight round through the revision fence, and the driver continues the new revision. The driver also stops on its own when a turn ends on max tokens, a durability write fails, the agent is cancelled, the plugin unloads, or the round cap is exhausted — at the cap it records a blocker with the stable code `round-limit`. Cancellation never auto-restarts a round: a goal whose round was under way is paused at the next idle point, and a cancellation unrelated to a goal attempt only disarms continuation. A round prompt still queued when the agent reaches idle is removed from the inbox, so later human input does not wait behind it; when the cancelled turn belonged to other work, continuation is disarmed rather than paused.
53
+ A round starts only at whole-agent idle and, when the subagent service is loaded, only after the agent's progressing descendants settle; completion, pause, and blocking suppress continuation; a host-initiated pause also aborts the turn already running, while a model-initiated pause inside its own turn finishes normally. An edit only invalidates an in-flight round through the revision fence, and the driver continues the new revision. The driver also stops on its own when a turn ends on max tokens, a durability write fails, the agent is cancelled, the plugin unloads, or the round cap is exhausted — at the cap it records a blocker with the stable code `round-limit`. Cancellation never auto-restarts a round: a goal whose round was under way is paused at the next idle point, and a cancellation unrelated to a goal attempt only disarms continuation. Agent disposal, such as a graceful process shutdown or a reload of the plugin that owns the Agent, aborts the round with the `disposed` cause and only disarms continuation, so the goal stays active, as on plugin unload; user, parent, and hook cancellations still pause it. A round prompt still queued when the agent reaches idle is removed from the inbox, so later human input does not wait behind it; when the cancelled turn belonged to other work, continuation is disarmed rather than paused.
54
54
 
55
55
  ### After resume, fork, or unload
56
56
 
package/README.zh.md CHANGED
@@ -50,7 +50,7 @@ kind: "package-reference"
50
50
 
51
51
  ### 何时停止续行
52
52
 
53
- Round 只在整个 agent 进入 idle 时启动;加载了子 agent 服务时,还要等该 agent 仍在推进的后代全部结束后才启动;完成、暂停和阻塞会阻止续行;宿主发起的暂停还会中止正在运行的轮次,而模型在自己轮次内发起的暂停会正常结束。编辑只会通过修订栅栏使进行中的 Round 失效,驱动器会继续新修订。驱动器也会在以下情况自行停止:轮次因 max tokens 结束、持久性写入失败、agent 被取消、插件卸载,或 Round 上限耗尽——上限耗尽时它会以稳定代码 `round-limit` 记录一个 blocker。取消绝不会自动重启 Round:Round 已在进行的 goal 会在下一次 idle 时被暂停;与 goal 尝试无关的取消只会停用续行。agent 进入 idle 时仍在排队的 Round 提示会从收件箱移除,后续人类输入不会排在它后面等待;被取消的轮次属于其他工作时,续行会被停用而不是暂停。
53
+ Round 只在整个 agent 进入 idle 时启动;加载了子 agent 服务时,还要等该 agent 仍在推进的后代全部结束后才启动;完成、暂停和阻塞会阻止续行;宿主发起的暂停还会中止正在运行的轮次,而模型在自己轮次内发起的暂停会正常结束。编辑只会通过修订栅栏使进行中的 Round 失效,驱动器会继续新修订。驱动器也会在以下情况自行停止:轮次因 max tokens 结束、持久性写入失败、agent 被取消、插件卸载,或 Round 上限耗尽——上限耗尽时它会以稳定代码 `round-limit` 记录一个 blocker。取消绝不会自动重启 Round:Round 已在进行的 goal 会在下一次 idle 时被暂停;与 goal 尝试无关的取消只会停用续行。Agent 被处置(例如进程正常关停,或重载拥有该 Agent 的插件)时,Round 以 `disposed` 原因中止,此时只停用续行,goal 保持 active,与插件卸载时一致;用户、父 Agent 与 hook 发起的取消仍会暂停它。agent 进入 idle 时仍在排队的 Round 提示会从收件箱移除,后续人类输入不会排在它后面等待;被取消的轮次属于其他工作时,续行会被停用而不是暂停。
54
54
 
55
55
  ### resume、fork 或卸载之后
56
56
 
package/lib/index.js CHANGED
@@ -276,6 +276,10 @@ function apply(ctx) {
276
276
  return;
277
277
  }
278
278
  if (event.data.reason.kind !== "aborted") return;
279
+ if (event.data.reason.reason.kind === "disposed") {
280
+ disarm(state);
281
+ return;
282
+ }
279
283
  if (state.attempt?.phase === "claimed" || state.attempt?.phase === "admitted") state.attempt.cancelled = true;
280
284
  else disarm(state);
281
285
  return;
package/locale/de.json ADDED
@@ -0,0 +1,6 @@
1
+ {
2
+ "meta": {
3
+ "title": "Automatische Fortsetzung des Ziels",
4
+ "description": "Lassen Sie die KI die Ziele der aktuellen Sitzung weiter vorantreiben, sofern die Bedingungen und Credits dies zulassen."
5
+ }
6
+ }
package/locale/es.json ADDED
@@ -0,0 +1,6 @@
1
+ {
2
+ "meta": {
3
+ "title": "Continuación automática del objetivo",
4
+ "description": "Deje que la IA continúe avanzando hacia los objetivos de la sesión actual según lo permitan las condiciones y los créditos."
5
+ }
6
+ }
package/locale/fr.json ADDED
@@ -0,0 +1,6 @@
1
+ {
2
+ "meta": {
3
+ "title": "Poursuite automatique de la cible",
4
+ "description": "Laissez l’IA continuer à faire avancer les objectifs de la session en cours dans la mesure où les conditions et les crédits le permettent."
5
+ }
6
+ }
package/locale/ja.json ADDED
@@ -0,0 +1,6 @@
1
+ {
2
+ "meta": {
3
+ "title": "目標の自動継続",
4
+ "description": "条件と予算が許す範囲で、AI が現在のセッションの目標を引き続き進められるようにします。"
5
+ }
6
+ }
package/locale/pt.json ADDED
@@ -0,0 +1,6 @@
1
+ {
2
+ "meta": {
3
+ "title": "Continuação automática desejada",
4
+ "description": "Deixe a IA continuar a avançar nos objetivos da sessão atual conforme as condições e os créditos permitirem."
5
+ }
6
+ }
package/locale/ru.json ADDED
@@ -0,0 +1,6 @@
1
+ {
2
+ "meta": {
3
+ "title": "Целевое автопродолжение",
4
+ "description": "Пусть ИИ продолжит продвигать цели текущей сессии, насколько позволяют условия и кредиты."
5
+ }
6
+ }
package/package.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "@deepseek-ai/dsh-goal-round-driver",
3
3
  "description": "Race-fenced same-session goal-round driver",
4
- "version": "0.2.1-rc.5",
4
+ "version": "0.2.2-alpha.3",
5
5
  "publishConfig": {
6
6
  "access": "public"
7
7
  },
@@ -29,25 +29,25 @@
29
29
  ],
30
30
  "license": "MIT",
31
31
  "peerDependencies": {
32
- "@deepseek-ai/dsh-agent": "0.2.1-rc.5",
33
- "@deepseek-ai/dsh-goal": "0.2.1-rc.5",
34
- "@deepseek-ai/dsh-llm": "0.2.1-rc.5",
35
- "@deepseek-ai/dsh-session": "0.2.1-rc.5",
36
- "@deepseek-ai/cordis": "~4.0.5",
37
- "@deepseek-ai/dsh-subagent": "0.2.1-rc.5"
32
+ "@deepseek-ai/dsh-agent": "0.2.2-alpha.3",
33
+ "@deepseek-ai/dsh-goal": "0.2.2-alpha.3",
34
+ "@deepseek-ai/dsh-llm": "0.2.2-alpha.3",
35
+ "@deepseek-ai/dsh-session": "0.2.2-alpha.3",
36
+ "@deepseek-ai/cordis": "~4.0.6-alpha.2",
37
+ "@deepseek-ai/dsh-subagent": "0.2.2-alpha.3"
38
38
  },
39
39
  "devDependencies": {
40
- "@deepseek-ai/dsh-agent": "0.2.1-rc.5",
41
- "@deepseek-ai/dsh-agent-loop": "0.2.1-rc.5",
42
- "@deepseek-ai/dsh-agent-loop-testkit": "0.2.1-rc.5",
43
- "@deepseek-ai/dsh-goal": "0.2.1-rc.5",
44
- "@deepseek-ai/dsh-llm": "0.2.1-rc.5",
45
- "@deepseek-ai/dsh-session": "0.2.1-rc.5",
46
- "@deepseek-ai/dsh-system-prompt": "0.2.1-rc.5",
47
- "@deepseek-ai/dsh-tools": "0.2.1-rc.5",
48
- "@deepseek-ai/cordis": "~4.0.5",
49
- "@deepseek-ai/dsh-session-projection": "0.2.1-rc.5",
50
- "@deepseek-ai/dsh-subagent": "0.2.1-rc.5"
40
+ "@deepseek-ai/dsh-agent": "0.2.2-alpha.3",
41
+ "@deepseek-ai/dsh-agent-loop": "0.2.2-alpha.3",
42
+ "@deepseek-ai/dsh-agent-loop-testkit": "0.2.2-alpha.3",
43
+ "@deepseek-ai/dsh-goal": "0.2.2-alpha.3",
44
+ "@deepseek-ai/dsh-llm": "0.2.2-alpha.3",
45
+ "@deepseek-ai/dsh-session": "0.2.2-alpha.3",
46
+ "@deepseek-ai/dsh-system-prompt": "0.2.2-alpha.3",
47
+ "@deepseek-ai/dsh-tools": "0.2.2-alpha.3",
48
+ "@deepseek-ai/cordis": "~4.0.6-alpha.2",
49
+ "@deepseek-ai/dsh-session-projection": "0.2.2-alpha.3",
50
+ "@deepseek-ai/dsh-subagent": "0.2.2-alpha.3"
51
51
  },
52
52
  "peerDependenciesMeta": {
53
53
  "@deepseek-ai/dsh-subagent": {