dsh-opencode-provider
Run the following command in DeepSeek Harness:
dsh plugin install goku54477/dsh-opencode-provider
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install goku54477/dsh-opencode-provider in your terminal to install this plugin; the source repository is https://github.com/goku54477/dsh-opencode-provider
About this plugin
DeepSeek Harness out of the box expects a DeepSeek API key, yet many developers already run a local OpenCode server with open-source models like mimo. dsh-opencode-provider bridges that gap: it serializes the full DSH conversation into a single prompt string, ships it to the local OpenCode server over plain HTTP, and translates the incoming SSE event stream into DSH StreamChunk values — all without a single API key. The adapter resolves the concurrent deadlock between SSE subscription and prompt delivery, handles both legacy and current OpenCode event envelopes, and gracefully skips unknown events. With a default base URL of 127.0.0.1:4096, it works out of the box.
Key capabilities include key-free local HTTP connectivity, token-by-token streaming text output, configurable timeouts and retry policies, clean SSE cancellation via abort signals, and dual-format SSE event parsing (session.text.delta and session.next.text.delta). Each stream() call spawns a fresh OpenCode session and sends a best-effort interrupt on completion, error, or abort.
This plugin is ideal for developers who already have a local OpenCode server, want to drive pure-text generation through DSH workflows, and do not need structured tool calling or multimodal inputs. It does not provide persistent cross-request sessions, image understanding, or model-initiated external tool calls — its focus is on smoothly embedding local OpenCode text inference into the Harness ecosystem.
Use Cases
- Connect a local OpenCode server to DeepSeek Harness workflows
- Run open-source model text inference without any API key
- Map OpenCode SSE event streams to DSH StreamChunk values
Best For
- Developers with a local OpenCode server already running
- Users seeking key-free pure-text generation through DeepSeek Harness
- Scenarios favoring local models without multimodal input or structured tool calling
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.