dominds 1.23.10 → 1.24.2

package/dist/docs/tellask-collab.md

@@ -53,9 +53,9 @@ Short version: a Tellask Session is a resumable thread, not autonomous backgroun
 For teammate Tellasks, the runtime lifecycle is:
 
 1. Tellask is emitted.
-2. Tellasker waits while Side Dialog runs.
-3. A response is supplied back.
-4. Tellasker resumes.
+2. The callee Side Dialog runs as background work; the caller may naturally idle or keep working from other concrete triggers.
+3. When the callee response is supplied back, that response becomes a new fact for the caller.
+4. The caller resumes only from that result-arrival fact, a queued prompt, user input, Diligence Push when no active callee is pending, or another explicit drive source.
 
 Critical operational fact:
 
@@ -66,7 +66,7 @@ So if more work is needed, the tellasker must issue the next Tellask explicitly.
 
 ### 2.4 Diligence Push boundary
 
-- Diligence Push helps the tellasker avoid going idle.
+- Diligence Push helps Main Dialogs avoid going idle when they have not reached a natural idle boundary such as a pending active callee.
 - It does not send teammate Tellasks on the agent’s behalf.
 - It is a pressure mechanism, not an execution orchestrator.
 

package/dist/docs/tellask-collab.zh.md

@@ -53,9 +53,9 @@
 对诉请者来说，一次队友诉请的运行时节奏是：
 
 1. 发出 `tellaskSessionless({ targetAgentId: "<teammate>", tellaskContent: "..." }) ...`。
-2. 当前对话进入挂起/等待（pending sideDialogs）。
-3. 被诉请者完成本轮驱动并回贴结果。
-4. 诉请者恢复继续。
+2. 被诉请者所在支线作为后台工作推进；诉请者可自然 idle，也可因其它明确驱动事实继续工作。
+3. 被诉请者完成本轮驱动并回贴结果；这条回贴成为诉请者的新事实。
+4. 诉请者只会因结果到达、queued prompt、用户输入、无 pending active callee 时的 Diligence Push 或其它显式驱动源继续。
 
 关键事实（与停滞问题直接相关）：
 
@@ -64,7 +64,7 @@
 
 ### 2.4 Diligence Push 的边界
 
-- Diligence Push 会鞭策诉请者继续思考和推进。
+- Diligence Push 会在主线没有 pending active callee 等自然 idle 边界时，鞭策主线对话继续思考和推进。
 - 它不负责自动补发队友诉请，也不改变 tellask 的调用生命周期。
 - 所以它能缓解“发呆”，但不能根治“口头安排不落地”的协作停滞。
 

package/dist/evt-registry.js

@@ -51,6 +51,7 @@ class DialogEventRegistryImpl {
             case 'new_q4h_asked':
             case 'q4h_answered':
             case 'sideDialog_created_evt':
+            case 'dlg_background_callee_summary_evt':
             case 'dlg_touched_evt':
                 break;
             default:

package/dist/llm/defaults.yaml

@@ -491,6 +491,7 @@ providers:
       kimi-for-coding:
         name: Kimi For Coding
         optimal_max_tokens: 180000
+        critical_max_tokens: 210000
         supports_thinking: true
         default_thinking: true
         supports_tool_choice: false
@@ -614,25 +615,27 @@ providers:
       kimi-k2.6:
         name: Kimi-K2.6
         optimal_max_tokens: 180000
+        critical_max_tokens: 210000
         supports_thinking: true
         default_thinking: true
         supports_tool_choice: false
         supports_image_input: true
-        context_length: 200000
-        input_length: 200000
-        output_length: 8192
-        context_window: '200K'
+        context_length: 262144
+        input_length: 262144
+        output_length: 32768
+        context_window: '262K'
       kimi-k2.5:
         name: Kimi-K2.5
         optimal_max_tokens: 180000
+        critical_max_tokens: 210000
         supports_thinking: true
-        default_thinking: false
+        default_thinking: true
         supports_tool_choice: false
         supports_image_input: true
-        context_length: 200000
-        input_length: 200000
-        output_length: 8192
-        context_window: '200K'
+        context_length: 262144
+        input_length: 262144
+        output_length: 32768
+        context_window: '262K'
   anthropic:
     name: Anthropic
     apiType: anthropic