DeepSeek Harness 插件

lory69060/cn-intel-mcp-dsh

Star 数 ★ 1 下载量(近 30 天) 603 分类 工具与能力 收录于 2026-08-25 npm cn-intel-mcp-dsh

中国硬科技供应链情报:33 条信号看板、中报验证追踪、命中率记录与问答,通过远程 MCP 接入(mcp__cn_intel__*)。

安装

# npm 包(预构建)

dsh plugin --profile web add cn-intel-mcp-dsh

# Release 预构建包

dsh plugin --profile web add "https://github.com/lory69060/cn-intel-mcp-dsh/releases/download/v0.1.0/cn-intel-mcp-dsh-0.1.0.tgz"

# GitHub 源码(首次需按提示配置 allowBuilds 构建授权后重试)

dsh plugin --profile web add github:lory69060/cn-intel-mcp-dsh

装任何插件都等于在你的机器上跑第三方代码,权限和你本人一样大——能读你的文件、用你的凭据、访问网络,工具审批管不到它。GitHub 来源的插件还会在安装时执行构建脚本——pnpm 默认拦截,所以安装可能停在 ERR_PNPM_GIT_DEP_PREPARE_NOT_ALLOWED 或 ERR_PNPM_IGNORED_BUILDS;dsh 会打印出需要添加的确切键名,把它加进该 profile 的 pnpm-workspace.yaml 的 allowBuilds 下,重跑一次即可装上。放行构建本身就是一次信任判断:请只安装可信来源,并尽量锁定 commit(github:owner/repo#sha)。

README

该插件的 README 只有英文版本。

China hard-tech supply chain intelligence plugin for DeepSeek Harness.

Connects DSH to the cn-intel MCP server (Streamable HTTP) and exposes 6 tools under mcp__cn_intel__*:

Tool What it gives the model
read_signal_board 33 structured information-gap signals (industry / predicted_on / verify_by / result)
read_earnings_tracker H1 2026 earnings forecast vs actual verification tracker
get_track_record Historical hit-rate statistics — the judgment layer, not raw data
ask_edge Natural-language Q&A over supply-chain information gaps
list_articles Research article index (China hard-tech supply chains)
read_article Full article text for citation-grounded analysis

Install

dsh plugin --profile web add cn-intel-mcp-dsh
# or from the Plugin Market: search "cn-intel"

Then set the token:

export CN_INTEL_MCP_TOKEN="<your-token>"

Trial token (shared, rate-limited to 200 req/day):

export CN_INTEL_MCP_TOKEN="cninte…ad3c"

Request a dedicated free token or Pro tier by opening an issue in this repo.

Verify

In the DSH tools panel you should see:

mcp__cn_intel__read_signal_board
mcp__cn_intel__read_earnings_tracker
mcp__cn_intel__get_track_record
mcp__cn_intel__ask_edge
mcp__cn_intel__list_articles
mcp__cn_intel__read_article

Example prompts

See examples/prompts.md for 5 ready-to-use prompts (compliance query, signal board, track record, earnings tracker, article reading).

How it works

The plugin is a thin dsh-mcp-client bridge to the remote cn-intel MCP endpoint (https://cn-intel-mcp.lory69060.workers.dev/mcp). Data lives on GitHub Pages (GEO-optimized); the Worker is stateless with 300s caching.

License

MIT

FAQ

Q: Do I need a DeepSeek API key for this plugin? A: Yes, DSH itself needs a model provider. The plugin only adds supply-chain data tools on top.

Q: Is the data free? A: The trial token gives 200 requests/day shared. Dedicated free tokens and a Pro tier are available on request via GitHub issues.

Q: What data sources back the signals? A: Public disclosures (earnings reports, exchange announcements) via akshare, cross-checked and timestamped. The track-record tool exposes historical hit rates so you can judge reliability yourself.

Q: Can I self-host? A: The data is public on GitHub Pages; the MCP endpoint is a thin proxy. See docs/architecture.md.

内容来自项目 README(GitHub)↗

链接

同类插件

查看整个分类 →

社区评论

评论公开保存在 GitHub Discussions。加载评论会连接 GitHub 和 Giscus;发表内容需要 GitHub 账号。