dsh-qwen38-local-qol
Qwen3.8 local inference QoL plugin for DSH with thinking budgets, smart compaction, and multi-engine support.
Install
$ dsh plugin --profile web add github:Yunado/dsh-qwen38-local-qol#v0.3.3Plugin Overview & Capabilities
AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.
Key Capabilities
- Per-request thinking budget control across llama.cpp, NInfer, TabbyAPI, and oMLX backends
- Tailored history compaction backend that trims prefill and disables thinking during checkpoint summarization
- Preservation of images in tool outputs using multimodal message schemas
- Interactive DSH settings tab providing real-time live configuration and independent memory per inference engine
- Preset declaration bundling the qwen38 agent preset directly into DSH profiles
Useful For
- Running local Qwen3.8 models on Apple Silicon (oMLX) or PC/Server (llama.cpp, NInfer, TabbyAPI) within DeepSeek Harness
- Controlling reasoning tokens dynamically per query across low, medium, and extra-high effort levels
- Preventing context truncation during long-running agent workflows via optimized compaction thresholds
Who It Fits
- Developers running local LLMs using DeepSeek Harness
- Users running Qwen3.8 models via llama.cpp, TabbyAPI, NInfer, or Apple Silicon MLX
- DSH power users needing granular context window and thinking budget management
Documented Limitations
- Vision token capacity projections on TabbyAPI and oMLX are not calibrated on pre-flight meters
- Multimodal tool outputs require backend server support for image_url content arrays
- Agent preset seams are available on web surfaces and do not mount in headless profiles
- Requires Node.js >=22.14 and compatible DSH host versions
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
package.json declares 0 runtime, 12 development, 6 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
GitHub reports the repository license as MIT.
A root package.json was captured and can be inspected by the registry.
Public GitHub source metadata is available for this registry snapshot.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- Yunado/dsh-qwen38-local-qol
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- e8b170d · 2026-10-17
This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.
Repository Activity
- GitHub stars
- 20GitHub stars
- Forks
- 2
- Open issues
- 0
- Last commit
- 2026-10-13
- Last release
- 2026-10-13
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Compatibility bridge running unmodified Pi ecosystem extensions natively on DeepSeek Harness.
View plugin →Real-time market research cockpit and MCP bundle for DeepSeek Harness financial and macro data.
View plugin →Full office suite integration for DeepSeek Harness supporting spreadsheets, docs, slides, bases, and canvases.
View plugin →DeepSeek Harness integration for ego-lite Chromium automation, providing 32+ structured ego_* tools and a live screencast watch panel with direct interaction.
View plugin →