Toukaiteio/dsh-effort-tweak
A DeepSeek Harness plugin that allows you to change the reasoning effort of custom models in WebUI.
Adds a small settings page for choosing the reasoning levels available to each model in a DeepSeek Harness custom provider — for the case where model capability must be edited in llm-pi-ai configuration but the official Web UI does not yet expose that control. Adds Settings → Reasoning effort to the DSH Web UI; lists models explicitly configured in llm-pi-ai.providers.<provider>.models; lets you choose off / minimal / low / medium / high / xhigh / max for one model at a time; preserves the model's other fields and existing provider-specific wire values; saves through DSH's settings service — no manual configuration-file editing needed.
Install
curl -fsSL https://raw.githubusercontent.com/Toukaiteio/dsh-effort-tweak/main/scripts/install-dsh-effort-tweak.sh | bashInstall script per README (English): curl -fsSL https://raw.githubusercontent.com/Toukaiteio/dsh-effort-tweak/main/scripts/install-dsh-effort-tweak.sh | bash. For another profile: download the script and run bash install-dsh-effort-tweak.sh --profile work --no-start. The script downloads the latest GitHub Release archive, verifies the GitHub-provided SHA-256 digest when available, keeps it under DSH_HOME/plugin-archives/dsh-effort-tweak/, installs it with dsh plugin, and starts the default web profile. Manual path: dsh plugin --profile web add ./dsh-effort-tweak-0.1.0.tgz.
Compatibility
DeepSeek Harness web profile. Custom providers configured in llm-pi-ai.
Details
- Repo: Toukaiteio/dsh-effort-tweak
- Category: Web UI Enhancements
- Stars: 3
- Version: GitHub Release archive (0.1.0 tag per README; no npm package)
- Last push: 2026-08-13
- First seen: 2026-08-13
Recent updates
Settings → Reasoning effort page; per-model reasoning levels (off→max); saves via settings service; no manual config edits.
FAQ
- What problem does it solve?
- Custom providers need reasoning-level capability edited in llm-pi-ai config, but the official Web UI doesn't expose that control yet.
- Which levels can I pick?
- off, minimal, low, medium, high, xhigh, or max — per model, one at a time.
- Does it edit config files?
- No — it saves through DSH's settings service and preserves the model's other fields.
Alternatives
JuneLearn/dsh-reasoning-settings · GuoMonth/dsh-multi-tenant · PerryLink/dsh-output-styles