cn-intel-mcp-dsh

Structured intelligence on China's hard-tech supply chains-semiconductors, new energy, high-end equipment-is scattered across earnings filings, exchange announcements, and broker reports. When you ask an AI assistant for analysis, it tends to return generic summaries without the specific, verifiable signals a real workflow demands. The cn-intel-mcp-dsh plugin closes that gap by bridging the cn-intel MCP source directly into DeepSeek Harness, so the model can pull structured information-gap signals, cross-reference data points, and cite primary sources right inside the conversation-no tab-switching, no manual spreadsheet assembly.

Six tools are exposed under the mcp__cn_intel__ namespace: a signal board carrying 33 information-gap entries tagged by industry, predicted date, and verification method; an earnings tracker pairing H1 2026 forecasts against actual results; a track-record endpoint that surfaces historical hit-rate statistics so users can judge reliability for themselves; a natural-language Q-and-A tool (ask_edge) for ad-hoc supply-chain questions; and list / read article tools for research-paper indexing and full-text citation. All underlying data derives from public disclosures, cross-checked via akshare and timestamped. The architecture is deliberately thin-a stateless Cloudflare Worker proxying to a GitHub Pages-hosted store with 300-second caching.

The plugin is aimed at equity researchers, compliance analysts, and industry strategists who follow China's hard-tech supply chains and want to fold primary-source signals into their daily AI-assisted workflow. A shared trial token allows 200 requests per day; dedicated free tokens and a Pro tier are available on request through the project's GitHub issues.