dsh-owui-chat2api
Run the following command in DeepSeek Harness:
dsh plugin install Wecury/dsh-owui-chat2api
Paste the following prompt into your AI chat to install this plugin:
Install the plugin in DeepSeek Harness Desktop by running dsh plugin install Wecury/dsh-owui-chat2api; source code: https://github.com/Wecury/dsh-owui-chat2api
About this plugin
Open WebUI aggregates a large pool of self-hosted and third-party models, but its API is not natively OpenAI /v1-compatible, so DeepSeek Harness Desktop cannot call those models out of the box. dsh-owui-chat2api fills that gap: it spins up a lightweight local reverse proxy that exposes your Open WebUI models through a standard OpenAI-compatible endpoint, removing the need to hand-roll an adapter.
The plugin also ships a full operations console inside the DSH web shell, covering one-click proxy start and stop, a Python and dependency health check, Open WebUI URL and port configuration, and a per-model usage dashboard (today, yesterday, month, all-time). For reasoning models that honour reasoning_effort, a single sync button probes each model for its supported levels and writes the result into the DSH configuration, so there is no manual editing required.
It is ideal for users who already self-host Open WebUI and want to plug multiple models seamlessly into DeepSeek Harness Desktop. All session data, the usage database, and proxy settings stay on this machine under the ~/.dsh directory, the proxy binds to 127.0.0.1 only, and plugin updates never wipe user data.
Use Cases
- Expose all Open WebUI models to DeepSeek Harness Desktop as a standard /v1 API
- Track per-model usage (today, month, total) directly from the DSH panel
- Auto-sync reasoning_effort levels when new inference models appear
Best For
- Self-hosters running Open WebUI with a growing set of models
- Developers who want local models in DSH without writing adapters
- Operators who need local usage auditing with zero data leaving the machine
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
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.