AI Agent Hub
Back to plugins
🤖

dsh-codex-relay

Model Inference Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install necokeine/dsh-codex-relay

Paste the following prompt into your AI chat to install this plugin:

In DeepSeek Harness, you can install this plugin by running the command 'dsh plugin install necokeine/dsh-codex-relay' or by visiting the source at https://github.com/necokeine/dsh-codex-relay.

About this plugin

When using DeepSeek Harness for model inference, developers often want to leverage Codex's powerful code generation and tool execution capabilities, but integrating it seamlessly without leaving the DSH ecosystem can be a practical challenge. The dsh-codex-relay plugin solves this by acting as a smart bridge between DSH and a local Codex instance, allowing Codex to serve as an inference backend effortlessly. Its core ability lies in delegating each model request to Codex's native agent loop via the local Codex app-server's JSON-RPC interface. This means that tool calls, command executions, and file modifications all happen within Codex's native environment, while DSH simply records the request and final answer without interfering with Codex's internal processes, thus preserving its integrity and efficiency. This plugin is ideal for developers, AI engineers, or researchers who already use DeepSeek Harness and Codex, seeking to harness Codex's code intelligence in a unified interface without switching tools or redundant setups. Whether for script writing, project debugging, or automation tasks, it provides a smooth experience, maximizing Codex's capabilities within the DSH workflow.

Use Cases

  • Generating and debugging code via Codex within DeepSeek Harness.
  • Executing automation scripts and tasks with Codex's native tools.
  • Managing model requests and Codex tool calls in a unified DSH interface.

Best For

  • Developers looking to seamlessly integrate Codex into DeepSeek Harness workflows.
  • AI engineers seeking efficient model inference backends for tool execution.
  • Researchers exploring AI agent applications in code generation.