Preface

When building analytical workflows with DeepSeek Harness (DSH), accessing China hard tech supply chain intelligence is often complicated by public data that is fragmented and lacks structured processing. The cn-intel-mcp-dsh plugin connects to the cn-intel MCP server and directly converts structured supply chain signals, earnings tracking, and article indexes into DSH-callable tools, eliminating the need to manually scrape and clean data.

Overview

This is a plugin for DeepSeek Harness, maintained by lory69060. It is a lightweight bridge that uses dsh-mcp-client to expose the remote cn-intel MCP service (Streamable HTTP) to DSH. The plugin itself does not provide model inference capabilities; it only adds a data retrieval layer, allowing large models to perform citation-supported analysis based on public disclosure data.

Core Features

The plugin exposes six tools in the DSH tool panel, primarily used for retrieving supply chain intelligence:

  • read_signal_board: Provides 33 structured information-gap signals, including industry, forecast time, verification time, and result fields.
  • read_earnings_tracker: H1 2026 earnings forecast and actual verification tracker.
  • get_track_record: Historical hit-rate statistics. This is assessment-layer data used to evaluate the reliability of information, not raw data.
  • ask_edge: A natural language Q&A interface for supply chain information gaps.
  • list_articles: Index of China hard tech supply chain research articles.
  • read_article: Retrieves full-text content for citation-supported analysis.

Installation and Enablement

  1. Run the installation command in the terminal:
    dsh plugin --profile web add cn-intel-mcp-dsh
  1. Set the authentication token. The plugin accesses MCP permissions through an environment variable:
    export CN_INTEL_MCP_TOKEN="<your-token>"
  1. Use the official trial token (shared, limited to 200 requests per day):
    export CN_INTEL_MCP_TOKEN="cninte…ad3c"

If you need a higher quota or a dedicated token, please apply via GitHub Issues.

Typical Usage

After successful installation, you should see six tools prefixed with mcp__cn_intel__ in the DSH tool panel. For specific prompt examples, refer to the examples/prompts.md file in the repository, which includes five ready-to-use prompts covering compliance queries, signal board analysis, track record review, earnings tracking, and article reading.

Applicable Scenarios and Notes

  • Target audience: DSH users and developers who need to access public China hard tech supply chain intelligence.
  • Permissions note: The plugin itself does not provide model inference capabilities. You must configure a DeepSeek API key first to run DSH.
  • Data source: Data is based on public disclosures from akshare (earnings reports, announcements, etc.) and is stored on GitHub Pages.
  • Service endpoint: The MCP proxy endpoint is https://cn-intel-mcp.lory69060.workers.dev/mcp. The server is stateless and has a 300-second caching mechanism.

Conclusion

cn-intel-mcp-dsh provides a structured China hard tech supply chain data interface for DeepSeek Harness. By connecting to cn-intel MCP, developers can directly query the signal board, earnings tracker, and article full text in their workflows, enabling end-to-end automation from data retrieval to citation-based analysis. The complete source code and documentation are available at the GitHub repository.