DeepSeek HarnessHub
提交插件

Yunado/dsh-qwen38-local-qol ↗★ 4

dsh-qwen38-local-qol

DeepSeek Harness QoL plugin for the local Qwen3.8 line (27B, Flash-Next): per-request thinking budgets on both llama.cpp and NInfer dialects, vision + tools, and a compaction backend whose summaries stop burning the output cap on thinking.

视觉
包名
dsh-qwen38-local-qol
版本
0.1.0
许可证
MIT
最近更新
2026年9月12日

安装

$npx -p @deepseek-ai/dsh dsh plugin --profile web add github:Yunado/dsh-qwen38-local-qol

说明文档

阅读完整 README ↗

dsh-qwen38-local-qol

English · 中文

A QoL plugin for DeepSeek Harness (DSH) for running Qwen3.8 locally (llama.cpp llama-server or NInfer — both serve the OpenAI-compatible /v1 API; at the config level also Qwen3.8-Flash-Next). No core patches, no pi-ai patchfile.

Install

dsh plugin --profile web add github:Yunado/dsh-qwen38-local-qol

Restart dsh web: at boot the plugin generates the qwen38 user preset from the standard preset's composition, and sets it as the default agent preset when no default is configured. New sessions use it automatically; existing sessions keep the preset they were created with.

What it supports

  • Per-request thinking budgets. The llama.cpp line sends the selected per-effort budget on every request (reasoning_effort + reasoning_budget_tokens — overrides the server's --reasoning-budget); the NInfer line has one global budget for all efforts (defaultThinkingBudget → the server's --default-thinking-budget).
  • A compaction backend. The summarizer's prefill is trimmed (recent reasoning only, images downgraded to text placeholders, tool results capped), and compaction calls run thinking-off at the line's full output cap — checkpoints stop getting truncated at the token cap.
  • A settings tab that configures both lines, live.

The settings tab

DSH settings → Qwen3.8 Local:

Per-line memory (connection, window numbers, budgets, trim knobs). The status dot is green when qwen38 is the default preset, amber when a different preset is the default, gray when the preset is missing. Changes apply live and persist to settings.yaml (hot-reloaded).

What can be tuned

The settings tab is the primary entry; headless profiles and patch/env accept the same fields (an id-scoped patch replaces the whole config; env covers what the patch does not set):

AreaFields (env vars)Defaults
ServerbaseURL, model, displayName, apiKey (DSH_QWEN38_BASE_URL / _MODEL / _DISPLAY_NAME / _API_KEY)http://localhost:8080/v1, qwen3.8-27b, same as model, none
Dialectdialect (DSH_QWEN38_DIALECT)llamacpp
WindowcontextWindow, maxTokens (DSH_QWEN38_CONTEXT_WINDOW / _MAX_TOKENS)229376, 24576
ThinkingthinkingBudgets (llamacpp, per effort), defaultThinkingBudget (ninfer, global), defaultEffort (DSH_QWEN38_DEFAULT_EFFORT){ low: 4096, medium: 8192, xhigh: 16384 }, 16384, medium
Prefill trimDSH_QWEN38_SUMMARIZE_IMAGES, DSH_QWEN38_SUMMARIZE_KEEP_TURNS, DSH_QWEN38_SUMMARIZE_TOOL_CHARS (env only)strip, 5, 2000

Limitations

  • Flash-Next is config-compatible, not artifact-verified (NInfer ships 27B NVFP4 only so far) — it runs the llama.cpp dialect with its own window/budget values.
  • The preset seam is a web-surface feature — headless profiles do not mount agent-presets rows; the provider route (thinking budgets) works on both surfaces.
  • Summarizer internals depend on the engine version — the wire rules (thinking off, full output cap) hold for every engine version; engine internals are outside the plugin's control.

Update

dsh plugin --profile web update dsh-qwen38-local-qol

github: dependencies resolve to an exact commit — if the profile lockfile still pins the commit first installed, remove and re-add the plugin to force re-resolution. Updates never touch the generated preset or the settings section.

Uninstall

  1. dsh plugin --profile web remove dsh-qwen38-local-qol
  2. Delete the qwen38 preset on the Agent presets page.
  3. Drop agent-presets: { default: qwen38 } from ~/.dsh/settings.yaml.
  4. Restart DSH.

Session history, transcripts, model lines and engines are not state the plugin owns.

Develop

pnpm install
pnpm test
pnpm run build:client   # rebuild lib/client.js after touching src/client*

Host half = plain ESM JavaScript with JSDoc; the browser half is built by scripts/build-client.mjs and shipped as the committed lib/client.js. Design details: DESIGN.md.

License

MIT

中文

给本地跑 Qwen3.8 的人用的 DeepSeek Harness（DSH）QoL 插件（llama.cpp llama-server 或 NInfer——两者都提供 OpenAI 兼容 /v1 API；配置层面兼容 Qwen3.8-Flash-Next）。零核心补丁、零 pi-ai 补丁文件。

安装

dsh plugin --profile web add github:Yunado/dsh-qwen38-local-qol

重启 dsh web：启动时插件从 standard preset 的组成生成 qwen38 用户 preset，且未配置默认时将其设为默认 agent preset。新会话自动使用；已有会话保留创建时的 preset。

功能特性

  • 逐请求 thinking 预算。 llama.cpp 线每请求发送所选 effort 的预算（reasoning_effort + reasoning_budget_tokens——覆盖服务端 --reasoning-budget）；NInfer 线全部 effort 共用一个全局预算（defaultThinkingBudget → 服务端 --default-thinking-budget）。
  • 压缩（compaction）后端。 摘要 prefill 先裁剪（只留近 N 轮 reasoning、图片降为文本占位符、工具结果按字数帽截断），且压缩调用强制 thinking off + 该线完整输出帽——checkpoint 不再被 token 帽截断。
  • 设置 tab：图形化配置两条线，即时生效。

设置 tab

DSH 设置 → Qwen3.8 本地：

按线记忆（连接、窗口数字、预算、裁剪旋钮）。状态圆点：绿 = qwen38 是默认 preset，黄 = 默认是别的 preset，灰 = preset 缺失。改动即时生效并持久化到 settings.yaml（热加载）。

可调项

设置 tab 是主入口；headless profile 或补丁/环境接受同样字段（按 id 定向的补丁替换整个 config，环境回退只覆盖补丁没写的字段）：

区域字段（环境变量）默认
服务器baseURL、model、displayName、apiKey（DSH_QWEN38_BASE_URL / _MODEL / _DISPLAY_NAME / _API_KEY）http://localhost:8080/v1、qwen3.8-27b、同 model、无
方言dialect（DSH_QWEN38_DIALECT）llamacpp
窗口contextWindow、maxTokens（DSH_QWEN38_CONTEXT_WINDOW / _MAX_TOKENS）229376、24576
ThinkingthinkingBudgets（llamacpp，按 effort）、defaultThinkingBudget（ninfer，全局）、defaultEffort（DSH_QWEN38_DEFAULT_EFFORT）{ low: 4096, medium: 8192, xhigh: 16384 }、16384、medium
Prefill 裁剪DSH_QWEN38_SUMMARIZE_IMAGES、DSH_QWEN38_SUMMARIZE_KEEP_TURNS、DSH_QWEN38_SUMMARIZE_TOOL_CHARS（仅环境变量）strip、5、2000

已知限制

  • Flash-Next 是配置兼容，未工件验证（NInfer 尚只出 27B NVFP4）——跑在 llama.cpp 方言，用自己的窗口/预算值。
  • preset 接缝是 web 面功能——headless profile 不挂 agent-presets 行；provider 路由（thinking 预算）两面都工作。
  • 摘要器内部行为依赖引擎版本——wire 层规则（thinking off、完整输出帽）对所有引擎版本生效；引擎内部行为不在本插件控制之内。

更新

dsh plugin --profile web update dsh-qwen38-local-qol

github: 依赖按精确 commit 解析——若 profile 锁文件仍钉在首次安装时的 commit，remove 后重新 add 插件即可强制重新解析。更新不触碰生成的 preset 与设置节。

卸载

  1. dsh plugin --profile web remove dsh-qwen38-local-qol
  2. 在 Agent 预设页删除 qwen38 preset。
  3. ~/.dsh/settings.yaml 删掉 agent-presets: { default: qwen38 }。
  4. 重启 DSH。

会话历史、transcript、模型线、引擎都不是插件持有的状态。

开发

pnpm install
pnpm test
pnpm run build:client   # 改完 src/client* 后重建 lib/client.js

宿主半边 = 带 JSDoc 的裸 ESM JavaScript；浏览器半边由 scripts/build-client.mjs 构建并以已提交的 lib/client.js 出货。设计细节：DESIGN.md。

许可

MIT