dsh-strata-sync 0.10.8 → 0.10.9

package/README.md

@@ -58,14 +58,16 @@ Existing conversations keep their selected model route and preset revision. To m
 | Setting | How it is chosen |
 |---|---|
 | Model and context | Strata's loaded text model and verified context capacity. |
-| Response output limit | Manual override, if set; otherwise the verified Strata output default. If unavailable or unlimited, use 30% of context, capped at 16,384 tokens. |
+| Response output limit | Manual override, if set; otherwise the verified Strata output default. If unavailable or unlimited, use Strata's formula: 30% of context, at least 4,096 and at most 32,768 tokens. |
 | Input types | Confirmed server capabilities, such as text or text plus image. |
 | Reasoning levels | Verified model-template support; unknown templates preserve existing settings. |
 | Default model | Follows the loaded Strata model when **Follow the loaded Strata model as the default** is enabled. |
 
 Keep **Follow Strata output limit** enabled for automatic selection. Turn it off to set a manual override under **Advanced settings**. The preview shows the server default, effective value and source separately.
 
-When synchronizing, the plugin sets a 4,096-token summary output cap and 4,096-token compaction headroom. The effective compaction trigger is the smaller of **80% of context** and **context − response output − 4,096**. Recent-content retention is **(context − response output) × 16%**, rounded down. These are client-side budgets calculated by the plugin, not settings supplied by Strata.
+The fallback follows [Strata's local output formula](https://github.com/Niko1221/Strata/blob/main/apps/host/server/agent/profile.mjs): `max(4096, min(32768, floor(context × 0.3)))`.
+
+When synchronizing, the plugin normally sets a 4,096-token summary output cap and compaction headroom. For small windows using calculated output, both are limited to half the remaining conversation budget, rounded down; an 8K window therefore uses 2,048. The effective compaction trigger is the smaller of **80% of context** and **context − response output − headroom**. Recent-content retention is **(context − response output) × 16%**, rounded down. These compaction budgets are calculated by the plugin, not supplied by Strata.
 
 For example, with automatic output calculation, a 32K context uses a 9,830-token response allowance and an 18,842-token compaction trigger. It does **not** necessarily compact at 80%.
 

package/dist/client.js

@@ -11,7 +11,7 @@ window.__ModuleLoader__.load({
 "Automatic calculation": {"zh-Hant":"自動計算","zh-Hans":"自动计算"},
 "Server has no fixed output limit.": {"zh-Hant":"伺服器未設定固定輸出上限。","zh-Hans":"服务器未设置固定输出上限。"},
 "Server output limit could not be verified.": {"zh-Hant":"無法確認伺服器的輸出上限。","zh-Hans":"无法确认服务器的输出上限。"},
-"Leave blank to use the server output default; if absent or unlimited, use 30% of context, up to 16,384. An explicit override takes priority.": {"zh-Hant":"留空會採用伺服器輸出預設值；未提供或無固定上限時，使用上下文的 30%，最多 16,384。明確填寫的覆寫值優先。","zh-Hans":"留空采用服务器输出默认值；未提供或无固定上限时，使用上下文的 30%，最多 16,384。明确填写的覆盖值优先。"},
+"Leave blank to use the server output default; if absent or unlimited, use 30% of context, at least 4,096 and at most 32,768. An explicit override takes priority.": {"zh-Hant":"留空會採用伺服器輸出預設值；未提供或無固定上限時，使用上下文的 30%，最低 4,096、最高 32,768。明確填寫的覆寫值優先。","zh-Hans":"留空采用服务器输出默认值；未提供或无固定上限时，使用上下文的 30%，最低 4,096、最高 32,768。明确填写的覆盖值优先。"},
 "Strata returned an invalid output limit. Review the server setting before syncing.": {"zh-Hant":"Strata 回傳無效的輸出上限，請先檢查伺服器設定。","zh-Hans":"Strata 返回无效的输出上限，请先检查服务器设置。"},
   "Runs once on the next DSH start; no periodic server checks.": {
     "zh-Hant": "下次啟動 DSH 時同步一次，不再定期檢查伺服器。",
@@ -377,7 +377,7 @@ window.__ModuleLoader__.load({
       'Enabled': {'zh-Hant':'已啟用','zh-Hans':'已启用'},
       'Save option changes below before checking differences again.': {'zh-Hant':'請先在下方儲存選項變更，再重新檢查差異。','zh-Hans':'请先在下方保存选项更改，再重新检查差异。'},
       'Follow Strata output limit': {'zh-Hant':'輸出上限跟隨 Strata','zh-Hans':'输出上限跟随 Strata'},
-      'Use the verified server default. If none is available, calculate 30% of context, up to 16,384 tokens.': {'zh-Hant':'採用已確認的伺服器預設值；無固定值時，以上下文的 30% 計算，最多 16,384 tokens。','zh-Hans':'采用已确认的服务器默认值；无固定值时，按上下文的 30% 计算，最多 16,384 tokens。'},
+      'Use the verified server default. If none is available, calculate 30% of context, at least 4,096 and at most 32,768 tokens.': {'zh-Hant':'採用已確認的伺服器預設值；無固定值時，以上下文的 30% 計算，最低 4,096、最高 32,768 tokens。','zh-Hans':'采用已确认的服务器默认值；无固定值时，按上下文的 30% 计算，最低 4,096、最高 32,768 tokens。'},
       'Advanced settings': {'zh-Hant':'進階設定','zh-Hans':'高级设置'},
       'Manual output limit': {'zh-Hant':'手動輸出上限','zh-Hans':'手动输出上限'},
       'Turn off Follow Strata output limit to enter a manual override.': {'zh-Hant':'關閉「輸出上限跟隨 Strata」後，可輸入手動覆寫值。','zh-Hans':'关闭“输出上限跟随 Strata”后，可输入手动覆盖值。'},
@@ -570,7 +570,7 @@ REASONING_DEFAULT_CONFLICT: 'The provider default reasoning level is not support
           h('summary',{className:'ss-option-title'},h('span',null,t('Sync preferences')),h('span',{className:'ss-option-description'},t('Output limit, default model and startup sync'))),
         h('div',{className:'ss-card'},
           h('label',{className:'ss-check'},h('input',{type:'checkbox',checked:!form.manual,'aria-label':t('Follow Strata output limit'),onChange:e=>edit({manual:!e.target.checked}),disabled:pending||!state}),t('Follow Strata output limit')),
-          h('p',{className:'ss-muted'},t('Use the verified server default. If none is available, calculate 30% of context, up to 16,384 tokens.')),
+          h('p',{className:'ss-muted'},t('Use the verified server default. If none is available, calculate 30% of context, at least 4,096 and at most 32,768 tokens.')),
           activeBudget?outputValues(activeBudget,preview.server?.model):h('p',{className:'ss-muted'},t('Check differences to read the server and effective limits.')),
           h('details',{className:'ss-details',open:form.manual},h('summary',null,t('Advanced settings')),
             h('label',null,t('Manual output limit'),h('input',{type:'number',min:1,step:1,value:form.output,onChange:e=>edit({output:e.target.value}),disabled:pending||!state||!form.manual})),

package/dist/core.mjs

@@ -59,18 +59,21 @@ export function normalizeBaseURL(value) {
 
 /**
  * Select the output allowance, then derive DSH's client-side compaction policy.
- * The 4096 headroom also matches the native pi-ai context safety margin.
+ * The usual 4096 headroom also matches the native pi-ai context safety margin.
  */
 export function planBudget(contextWindow, { outputTokens, serverOutputTokens } = {}) {
   if (!isContext(contextWindow) || contextWindow < 8192) {
     fail('UNSUPPORTED_CONTEXT', 'A verified context window of at least 8,192 tokens is required.');
   }
-  const output = outputTokens ?? serverOutputTokens ?? Math.min(16384, Math.floor(contextWindow * 0.3));
+  const output = outputTokens ?? serverOutputTokens ?? Math.max(4096, Math.min(32768, Math.floor(contextWindow * 0.3)));
   const outputSource = outputTokens != null ? 'manual' : serverOutputTokens != null ? 'server' : 'calculated';
   if (!Number.isSafeInteger(output) || output < 1 || output >= contextWindow) {
     fail('INVALID_OUTPUT_BUDGET', 'The output budget must be a positive integer smaller than the context window.');
   }
-  const headroomTokens = 4096;
+  // The 4K output floor must still leave compaction room in an 8K window.
+  // Keep fixed/manual budgets and the usual 4K reserve unchanged.
+  const headroomTokens = outputSource === 'calculated'
+    ? Math.min(4096, Math.floor((contextWindow - output) / 2)) : 4096;
   const thresholdRatio = 0.8;
   const retainRatio = 0.16;
   const messageBudgetTokens = contextWindow - output;

package/docs/README.zh-CN.md

@@ -58,14 +58,16 @@ dsh plugin --profile desktop add dsh-strata-sync --registry=https://registry.npm
 | 设置 | 如何决定 |
 |---|---|
 | 模型与上下文 | 采用 Strata 已加载的文本模型及已确认的上下文容量。 |
-| 回复输出上限 | 优先使用手动覆盖值；否则采用已确认的 Strata 输出默认值。无法取得或未限制时，使用上下文的 30%，最多 16,384 tokens。 |
+| 回复输出上限 | 优先使用手动覆盖值；否则采用已确认的 Strata 输出默认值。无法取得或未限制时，使用 Strata 官方公式：上下文的 30%，最低 4,096、最高 32,768 tokens。 |
 | 输入类型 | 采用服务器已确认的能力，例如文本或文本＋图片。 |
 | 推理等级 | 根据已验证的模型模板确定；未知模板保留原有设置。 |
 | 默认模型 | 启用**默认模型跟随 Strata 已加载的模型**时，跟随服务器当前模型。 |
 
 保持**输出上限跟随 Strata**开启，即可自动选择。关闭后可在**高级设置**输入手动覆盖值。预览会分别显示服务器默认值、实际应用值及其来源。
 
-同步时，插件将摘要输出上限和压缩预留空间均设为 4,096 tokens。实际压缩阈值取**上下文的 80%**与**上下文 − 回复输出预算 − 4,096**中的较小值，近期内容保留量为 **（上下文 − 回复输出预算）× 16%**，向下取整。这些是插件计算的客户端预算，并非 Strata 提供的设置。
+自动计算采用 [Strata 官方输出公式](https://github.com/Niko1221/Strata/blob/main/apps/host/server/agent/profile.mjs)：`max(4096, min(32768, floor(context × 0.3)))`。
+
+同步时，摘要输出上限和压缩余量通常均为 4,096 tokens。小窗口使用自动计算输出时，两者最多为剩余对话预算的一半，向下取整；因此 8K 窗口使用 2,048。实际压缩阈值取**上下文的 80%**与**上下文 − 回复输出预算 − 压缩余量**中的较小值，近期内容保留量为 **（上下文 − 回复输出预算）× 16%**，向下取整。这些压缩预算由插件计算，并非 Strata 提供的设置。
 
 例如，采用自动输出计算时，32K 上下文的回复预算为 9,830 tokens，压缩阈值为 18,842 tokens，**不一定在 80% 时压缩**。
 

package/docs/README.zh-TW.md

@@ -58,14 +58,16 @@ dsh plugin --profile desktop add dsh-strata-sync --registry=https://registry.npm
 | 設定 | 如何決定 |
 |---|---|
 | 模型與上下文 | 採用 Strata 已載入的文字模型及已確認的上下文容量。 |
-| 回覆輸出上限 | 優先使用手動覆寫值；否則採用已確認的 Strata 輸出預設值。無法取得或未限制時，使用上下文的 30%，最多 16,384 tokens。 |
+| 回覆輸出上限 | 優先使用手動覆寫值；否則採用已確認的 Strata 輸出預設值。無法取得或未限制時，使用 Strata 官方公式：上下文的 30%，最低 4,096、最高 32,768 tokens。 |
 | 輸入類型 | 採用伺服器已確認的能力，例如文字或文字＋圖片。 |
 | 推理等級 | 依已驗證的模型範本決定；未知範本保留原有設定。 |
 | 預設模型 | 啟用**預設模型跟隨 Strata 已載入的模型**時，跟隨伺服器目前模型。 |
 
 保持**輸出上限跟隨 Strata**開啟，即可自動選擇。關閉後可在**進階設定**輸入手動覆寫值。預覽會分別顯示伺服器預設值、實際套用值及其來源。
 
-同步時，外掛將摘要輸出上限和壓縮預留空間均設為 4,096 tokens。實際壓縮門檻取**上下文的 80%**與**上下文 − 回覆輸出預算 − 4,096**中的較小值，近期內容保留量為 **（上下文 − 回覆輸出預算）× 16%**，向下取整。這些是外掛計算的客戶端預算，並非 Strata 提供的設定。
+自動計算採用 [Strata 官方輸出公式](https://github.com/Niko1221/Strata/blob/main/apps/host/server/agent/profile.mjs)：`max(4096, min(32768, floor(context × 0.3)))`。
+
+同步時，摘要輸出上限和壓縮餘裕通常均為 4,096 tokens。小視窗使用自動計算輸出時，兩者最多為剩餘對話預算的一半，向下取整；因此 8K 視窗使用 2,048。實際壓縮門檻取**上下文的 80%**與**上下文 − 回覆輸出預算 − 壓縮餘裕**中的較小值，近期內容保留量為 **（上下文 − 回覆輸出預算）× 16%**，向下取整。這些壓縮預算由外掛計算，並非 Strata 提供的設定。
 
 例如，採用自動輸出計算時，32K 上下文的回覆預算為 9,830 tokens，壓縮門檻為 18,842 tokens，**不一定在 80% 時壓縮**。
 

package/docs/VALIDATION.md

@@ -11,7 +11,9 @@ Validation date: 2026-10-30 (Asia/Taipei).
 
 ## Automated verification
 
-All **160 automated tests pass** for the current runtime. Coverage includes:
+All **162 automated tests pass** for the current runtime. Coverage includes:
+
+- Automatic output matches Strata's 30% formula with a 4,096 floor and 32,768 ceiling. Tests cover 8K through 1M contexts, rounding at the ceiling, unchanged server/manual precedence, valid small-window compaction budgets and all three UI translations. The 8K calculated-output case limits compaction headroom and summary cap to 2,048; 32K and larger retain 4,096.
 
 - Actual pi-ai context clamping across multiple input sizes, near-full and exhausted context, independent source length, additional prompt costs, disabled-mode zero discovery and restored native requests.
 - Native Cordis waterfall dispatch, immutable messages, concurrent summaries, duplicate hooks, cancellation, metadata failures, early stream closure, native errors and unload.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-strata-sync",
-  "version": "0.10.8",
+  "version": "0.10.9",
   "description": "Native DSH settings UI for synchronizing Strata Server models, context and compaction budgets",
   "type": "module",
   "main": "dist/index.js",