use-opencode-local-provider
Run the following command in DeepSeek Harness:
dsh plugin install Payel-git-ol/use-opencode-local-provider
Paste the following prompt into your AI chat to install this plugin:
In DeepSeek Harness, run dsh plugin install Payel-git-ol/use-opencode-local-provider to install this plugin. The full source code is available at https://github.com/Payel-git-ol/use-opencode-local-provider .
About this plugin
OpenCode Zen models are capable, yet dsh does not expose them as a native provider out of the box, so wiring them up usually means rolling your own forwarding layer. This plugin removes that friction: install it and the full OpenCode Zen catalog appears in the dsh chat UI immediately.
Under the hood, the plugin ensures an opencode serve instance is running on load, then spins up a lightweight OpenAI-compatible bridge that maps /v1/chat/completions requests into opencode sessions over the local HTTP API. The flow travels through the opencode client channel, meaning no OpenCode API key is needed and public /zen/v1 quota stays untouched. Crucially, the bridge preserves opencode's native tool calls and connected MCP servers, letting the model decide when to invoke them inside its own sandbox, while dsh simply receives the final chat completion and stays out of the tool-orchestration loop.
Ideal for developers who already have opencode installed and want to drop the OpenCode Zen model family directly into their dsh interface without extra API keys or public quota consumption. Configuration covers server and bridge ports, working directory, stream timeout, and an automatic permission-reply policy (once, always, or reject), so sensible defaults get you running right out of the box.
Use Cases
- Use the full OpenCode Zen model catalog directly in the dsh chat UI for inference
- Invoke opencode tools and connected MCP servers through the local bridge for multi-step tasks
- Run OpenCode Zen locally without an API key or public quota consumption
Best For
- Developers who have opencode installed and want its models available in dsh
- Technical users needing local inference without relying on public API quota
- AI workflow builders who want to manage multiple model providers uniformly in dsh
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.