dsh-toolcall-stream-repair
Run the following command in DeepSeek Harness:
dsh plugin install udsy19/dsh-toolcall-stream-repair
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install udsy19/dsh-toolcall-stream-repair inside DeepSeek Harness to install this plugin; source code is available at https://github.com/udsy19/dsh-toolcall-stream-repair
About this plugin
When running model inference in DeepSeek Harness, streaming tool-call outputs can occasionally be truncated, malformed, or lost mid-stream, breaking downstream parsing and interrupting the Agent workflow. dsh-toolcall-stream-repair is a lightweight repair plugin designed specifically for this pain point.
The plugin performs real-time repair and completion of tool-call streaming responses within the inference pipeline. Even when the raw stream is incomplete or structurally inconsistent, developers still receive a correctly parseable, well-formed tool-call result, reducing reliance on the upstream model to produce perfectly structured output and improving the robustness of the entire chain.
If you are building multi-turn tool-calling Agent applications in DeepSeek Harness, or you keep hitting truncated, missing-field, or malformed tool-call streams under streaming inference, this plugin is worth adding to your stack.
Use Cases
- Multi-turn tool-calling Agent development where streaming responses are truncated or malformed
- Streaming inference scenarios requiring guaranteed integrity and parseability of tool-call outputs
- Reducing reliance on upstream model output perfection to improve pipeline robustness
Best For
- Developers building Agent applications in DeepSeek Harness
- Inference pipeline engineers frequently handling streaming tool-call anomalies
- Teams pursuing high availability and fault tolerance in tool-calling workflows
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.