luxueliu-reasoning-efforts
Run the following command in DeepSeek Harness:
dsh plugin install luxueliu/luxueliu-reasoning-efforts
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install luxueliu/luxueliu-reasoning-efforts in your DeepSeek Harness terminal to install. Source: https://github.com/luxueliu/luxueliu-reasoning-efforts
About this plugin
DeepSeek Harness has a known limitation: the reasoning-effort selector only appears for models declared as reasoning-capable, which in practice means the DS family. Models routed through a local LiteLLM gateway such as Gemini, Grok, Kimi, and GLM have no slider at all to toggle thinking depth in conversation.
This plugin fills that gap with three layers of capability. An efforts.json truth table ships with 20 pre-set gateway aliases, but slots are unlimited so you can add as many models as your gateway serves. A /efforts slash-command diffs your live settings against the gold sample, detects drift, and re-applies effort levels with automatic backup. An optional upstream patch for llm-pi-ai introduces a gateway-level default so every manually declared model inherits the table from a single declaration.
If you route multiple vendor models through LiteLLM or a similar gateway into DSH and are tired of hand-editing settings.yaml after every upgrade, this plugin gives your non-DS models the same off / low / medium / high / xhigh / max slider experience as the native ones.
Use Cases
- Show reasoning-effort sliders for multi-vendor models routed through a local gateway into DSH
- Detect and reapplies drifted effort levels in one click after a DSH upgrade
- Bulk-declare reasoning tiers for any number of custom gateway models
Best For
- Multi-model users routing non-DS models into DSH via a LiteLLM gateway
- DSH users who find hand-editing settings.yaml after every upgrade tedious
- Gateway operators wanting to manage reasoning tiers across many models in one place
Related Plugins
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.