Agent & Workflowdsh-plugin
Stars
15
Forks
1
Open issues
1
Last push
Oct 21, 2026
Latest release
v0.36.8
24h Growth
+0
7d Growth
+3
Installation
dsh plugin --profile web add https://github.com/TikaFlow/dsh-model-fix/releases/download/v0.36.8/dsh-model-fix.tgzOverview
dsh-model-fix is a DeepSeek Harness (DSH) plugin that provides model parameter filling from models.dev data, per-model reasoning-level memory, compatibility options, subagent following of the parent agent, AI self-tuning of reasoning effort, and visual settings.
- Auto-fills missing model parameters from models.dev data, including reasoning level, context window, max output, and image modality.
- Allows updating existing model parameters when enabled; default configuration leaves existing values unaffected.
- Compatibility options can disable the developer role and remove the store parameter for endpoints that do not accept them.
- Remembers the last manually selected reasoning level for each model and restores it when switching models.
- Uses high as the default reasoning level when no memory exists and the model offers a high level.
- Subagents can follow the parent agent's provider, model, and reasoning level in real time.
- AI can autonomously adjust the reasoning level according to task difficulty.
- Provides visual settings, model validation, and probing of supported reasoning levels.