dsh-skill-zh-distribute
v0.1.0
Published
DeepSeek Harness skill — adapt one piece of source content (母本) into platform-native posts for 掘金/微信公众号/X/小红书/B站 and schedule them on a staggered, anti-spam-friendly cadence. Pure methodology + per-platform dossiers, zero runtime code; honest that auto-po
Maintainers
Readme
dsh-skill-zh-distribute
A DeepSeek Harness skill that turns one piece of source content into platform-native posts for the five main Chinese-distribution surfaces — 掘金 / 微信公众号 / X / 小红书 / B站 — on a staggered, anti-spam schedule.
What it is
dsh-skill-zh-distribute is a pure-guidance skill for DeepSeek Harness (DSH). It is a publishing router: one structured 母本 (source block) → one copy-paste-ready post per platform, each in that platform's native "头部博主" format, plus a staggered posting schedule.
It is honest first: auto-posting is blocked on most Chinese surfaces (X Free = 402 credits; 公众号/小红书/B站 have no free write API). The real zero-cost model is AI generates the per-platform copy → one daily-pack-DATE.md → a human pastes each block in ~5 minutes. This skill makes that 5-minute paste as effective as possible.
What it gives you: the 母本 source-block schema; per-platform dossiers (掘金/公众号/X/小红书/B站: audience, format, length, tone, tags, CTA, gotchas); the adaptation call shape; the one-pack output format; and a staggered schedule with anti-spam reasoning.
What it does not do: post for you, bypass anti-spam, make X Free-tier credits appear, or guarantee reach.
Install (zero-code skill)
DSH discovers skills from the filesystem skill roots (~/.dsh/skills/, <project>/.dsh/skills/, ~/.agents/skills/), one level deep: <root>/<name>/SKILL.md. The bundled installer links SKILL.md into ~/.dsh/skills/zh-distribute/:
npx dsh-skill-zh-distribute # one command (recommended)
# or
npm i -g dsh-skill-zh-distribute && dsh-skill-zh-distribute
# or manual
mkdir -p ~/.dsh/skills/zh-distribute && cp SKILL.md ~/.dsh/skills/zh-distribute/SKILL.mdRestart DSH (or let its watcher pick it up). The skill zh-distribute appears in the catalog; the model loads it on demand.
dsh plugin addinstalls Cordis bundles (code plugins). A zero-code skill has no bundle, so it loads via the skill filesystem provider — which is what the linker above does.
Usage
In a DSH conversation, describe the distribution goal; the catalog's "Use when…" routes this skill:
- "把这篇 GitHub 精选分发到掘金/公众号/X/小红书/B站"
- "做一个每日分发包,5 分钟发完"
- "X thread / 小红书 / 公众号 各自怎么写?"
- "错峰排期怎么排?为什么不能同一分钟发五个平台?"
Example pack
# 📦 一键分发包 2026-10-03 (母本日期 2026-10-03)
> 5 平台 · 各按头部博主格式 · 5 分钟发完
## ── 1. 掘金(复制即发)──
### GitHub每日精选 | 今日3个值得关注的开源项目
...
## ── 2. 微信公众号(复制即发)──
今日GitHub热门:这3个AI项目值得收藏
...
## ── 3. X / Twitter(复制即发)──
🔥 3 个今天挖到的 AI 宝藏
---
repo名 · 一句话价值 · 链接
...
Follow @<handle> 不迷路
## ── 4. 小红书(复制即发)──
【3个AI神器】程序员效率翻倍
1️⃣ ... 2️⃣ ... 3️⃣ ...
点❤️收藏不迷路✨ #GitHub #开源 ...
## ── 5. B站(复制即发)──
动态:今天3个宝藏项目👇 ...Honest limits
- No free auto-posting on X/公众号/小红书/B站. AI copy + manual paste is the zero-cost path; paid official APIs are a later-stage decision.
- External links are mostly dead on CN surfaces (公众号/小红书 suppress them). Design each post to be worth saving even without a clickable link.
- Reach is not guaranteed — format fidelity raises the floor, traffic compounds over weeks.
- The model hallucinates — read each variant before pasting; manual paste is your last defense.
- Don't spam — daily cadence, not flood; respect each platform's volume norms.
License
MIT.
这是什么
dsh-skill-zh-distribute 是 DeepSeek Harness (DSH) 的纯指导 skill,一个发布路由器:一条结构化母本 → 每个平台一条"复制即发"的本土化帖子(掘金/公众号/X/小红书/B站),外加错峰排期。
先说大实话:中文分发平台大多不能免费自动发(X Free = 402 credits;公众号/小红书/B站无免费写 API)。真正零成本的模型是 AI 生成各平台文案 → 存一个 daily-pack-DATE.md → 人 5 分钟复制粘贴。本 skill 就是让这 5 分钟粘贴尽量高效。
给你的:母本 schema、五平台 dossier(受众/格式/长度/口吻/标签/CTA/坑)、适配调用形状、分发包输出格式、错峰排期+反垃圾理由。不给你:替你发、绕过反垃圾、凭空给 X Free 额度、保证流量。
安装(零代码 skill)
DSH 从文件系统 skill 根(~/.dsh/skills/、<项目>/.dsh/skills/、~/.agents/skills/)一层深发现 <root>/<name>/SKILL.md。本包带一键安装器:
npx dsh-skill-zh-distribute # 推荐:一键链接
# 或
npm i -g dsh-skill-zh-distribute && dsh-skill-zh-distribute
# 或手动
mkdir -p ~/.dsh/skills/zh-distribute && cp SKILL.md ~/.dsh/skills/zh-distribute/SKILL.md链好后重启 DSH,zh-distribute 出现在目录里,模型按需加载。
dsh plugin add装的是 bundle(代码插件);零代码 skill 走 skill 文件系统 provider,用上面的链接器。
用法
在 DSH 对话里直接描述分发目标,目录里的 "Use when…" 会自动路由:
- "把这篇 GitHub 精选分发到掘金/公众号/X/小红书/B站"
- "做一个每日分发包,5 分钟发完"
- "X thread / 小红书 / 公众号 各自怎么写?"
- "错峰排期怎么排?为什么不能同一分钟发五个平台?"
License
MIT。
