In the plugin ecosystem of DeepSeek Harness (DSH), code review often needs to be integrated into agent workflows. Existing solutions may not be structured enough to directly output scores. xby-review-code is a code review tool server based on the Model Context Protocol (MCP), designed to meet the needs of code review and scoring in a DSH environment.

Plugin Introduction

This is a code review tool server based on the Model Context Protocol (MCP), part of the DSH plugin system. It does not directly call large language models; instead, it constructs review prompts and provides score parsing capabilities.

Core Capabilities

The plugin provides the following five core capabilities:

  • set_xby_apikey: Sets the API key in the chat. It supports automatic persistence and remains valid after restart.
  • review_code: Constructs an LLM prompt for overall code review and scoring.
  • review_diff: Constructs an LLM prompt for reviewing and scoring Git diff changes.
  • review_file: Constructs an LLM prompt for single-file review and scoring.
  • parse_review_score: Extracts score information from review text (format: Total score: XX points).

Installation and Enablement

Prerequisites

Before using this plugin, you need to register on the official XiaoBenYang website and obtain an API key.

Installation Command

Run the following command in the terminal to install the plugin:

dsh plugin --profile web add github:xby_skill/xby-review-code

After installation, invoke the set_xby_apikey command in the DSH chat interface with the key obtained from the official XiaoBenYang website to complete the configuration.

Typical Usage

  1. Set the key: Enter set_xby_apikey <your key> in the conversation. The system will automatically save it.
  2. Construct the prompt: Depending on the review target, call review_code, review_diff, or review_file. These functions return the prompt to be used for LLM review, not the direct review result.
  3. Parse the score: Pass the review text returned by the LLM to parse_review_score to extract the total score.

Notes

  • No direct LLM calls: The plugin itself does not directly call LLMs; it only constructs prompts. Actual review requires integrating with DSH context or making subsequent large language model calls.
  • Depends on an external service: You must obtain the API key from the official XiaoBenYang website and set it in the DSH session.
  • Permissions and security: The plugin runs with the permissions of the current dsh process. It is recommended to review the source code and license before installation.

Summary

This tool provides a standardized entry point for code review in DSH workflows. For more details and the source code, see GitHub.