lhmd/dsh-promotion-toolkit
Turn any idea into platform-native publicity
DSH Promotion Toolkit is a DeepSeek Harness native plugin that turns any idea — product, open-source project, article, course, event, personal opinion, or a rough draft — into publishable, platform-native publicity content. It first extracts core theses, citable evidence, risks needing verification, and original links, then reshapes the content for each platform's native structure instead of shortening the same paragraph 17 times: Xiaohongshu checklist posts, Zhihu Markdown explainers, WeChat long-form, Moments friend-style shorts, Bilibili/Douyin/Kuaishou shot lists and scripts, Weibo/X/Threads hot takes with questions, LinkedIn career context, Reddit community framing, TikTok/YouTube video scripts, Medium long-form, and Facebook/Instagram shares. Numbers, quotes, dates, and effects are never silently filled in — the plugin flags what needs human verification against the source.
Install
npm install -g @deepseek-ai/dsh && dsh plugin --profile web add /absolute/path/to/dsh-promotion-toolkitLocal-path profile bundle install per README (English edition; source checkout required, no npm package documented): npm install -g @deepseek-ai/dsh, dsh plugin --profile web add /absolute/path/to/dsh-promotion-toolkit, dsh web. The repo follows the DSH profile-bundle shape: cordis.patch.yml mounts the package; src/index.js exports apply(ctx) and registers the viral_kit tool; src/skill.js provides the runtime Skill; skills/viral-kit/SKILL.md is the readable mirror. CLI for headless runs: node scripts/viral-kit.mjs --text "..." --language auto --platform linkedin --source-url https://github.com/lhmd.
Compatibility
DeepSeek Harness profile-bundle plugin. Registers the viral_kit tool that turns a product story, open-source project, article, course, event, or personal idea into platform-native publicity: it extracts the thesis, evidence, risk flags, and canonical URL first, then changes the reading path for each surface. 17 platforms: xiaohongshu, zhihu, wechat, wechat_moments, weibo, bilibili, douyin, kuaishou, x, threads, linkedin, reddit, tiktok, youtube, medium, facebook, instagram. It will not invent numbers, quotes, dates, or effects, and reminds you to verify against the source before publishing. MIT.
Details
- Repo: lhmd/dsh-promotion-toolkit
- Category: Coding & Development
- Stars: 9
- Version: Local-path install /absolute/path/to/dsh-promotion-toolkit (README-documented; no npm package)
- Last push: 2026-08-13
- First seen: 2026-08-13
Recent updates
The current English README (README.en.md) documents: the problem it solves, the platform differentiation table, 17 supported platforms, real deepseek-v4-pro case runs (4 long-form topics, Chinese and English, 17 platform versions each), install into DeepSeek Harness, the CLI, and development commands (npm test, check:release, demo).
FAQ
- Does it just shorten the same text 17 times?
- No — it first extracts the core thesis, citable evidence, risk boundaries, and canonical URL, then changes the reading path for each platform so the same facts are structured natively per surface (checklist, explainer, script, thread, etc.).
- Will it invent facts?
- No — numbers, quotes, dates, and effects are not silently added; the toolkit extracts only what is in the source and flags what needs human verification before publishing.
- Which platforms are covered?
- 17: xiaohongshu, zhihu, wechat, wechat_moments, weibo, bilibili, douyin, kuaishou, x, threads, linkedin, reddit, tiktok, youtube, medium, facebook, instagram.
Alternatives
lhmd/dsh-director-toolkit · CanglongCl/dsh-web-review · omdsh-dev/dsh-advisor