Dsh Better Reasoning Effort
HaoyueQin/dsh-better-reasoning-effort · skills
Reasoning-effort editing for third-party models in DeepSeek Harness: per-model thinking levels with a knowledge base + protocol inference, edited inside the official Models page card.
★ 18GitHub stars
0Forks
2026-09-08Last updated
TypeScriptLanguage
MITLicense
Key features
- In-page injectionan editor block appears in the official Models page under each model row's disclosure, next to context window / max tokens — not a separate settings page, but part of the official editing flow (same settings.mutate contract, same save style).
- Only the official DeepSeek API (the built-in catalog) can set reasoning effort
- Knowledge base + protocol inference: suggestEfforts() in src/knowledge.ts, a pure function shared by host and browser — fusing endpoint signals, curated entries (levels, modalities, reference capacities), a name heuristic, and protocol inference.
- the session model-selection contract it rides (per-session model directory → adapter-advertised effort ladder → selectModel submit, with optimistic snap and rollback on refusal)
- The block spans the full row
Requirements
- Injection depends on the official Models page's current DOM (aria-label/class).
- If an official upgrade changes the structure, injection pauses until adapted
Install command
dsh plugin --profile web add github:HaoyueQin/dsh-better-reasoning-effort