Multi-agent deep long-form article pipeline skill for DeepSeek Harness, running 7 roles across 5 phases with parallel literature/data/case retrieval, an evidence base, independent audit and human verification gates.
Install
# from GitHub (first run asks for allowBuilds approval — follow the hint, retry)
dsh plugin --profile web add github:zuoyunlai/lunheng-article-pipeline-dsh
Any plugin you install runs third-party code with your own permissions — it can read your files, use your credentials, and reach the network, and tool approvals don’t sandbox it. GitHub-sourced plugins also run build scripts at install time. Only install sources you trust, and pin a commit (github:owner/repo#sha).
README
This plugin publishes its README in Chinese only.
多 Agent 深度长文流水线技能包,DeepSeek Harness(dsh)bundle 插件版。
把一篇深度文章/论文的生产拆成 7 个角色 + 5 个阶段(Phase 1 为 T1 文献∥T2 数据∥T6 案例 三检索员三方真并行、互不干涉,T6 任何量级必 spawn 含 0 条场景空卡协议),用 dsh subagent 子代理编排,产出有证据底座、反方论证、独立审计、人工核验节点的交付物。
版本:v2.1.8-dsh.1(DSH 适配版,对应正典 v2.1.8)。
安装(在目标机器上)
# 1) 装进 profile 的 node_modules
dsh plugin --profile web add lunheng-article-pipeline
# 2) 把 bundle 加入 profile 清单
# $DSH_HOME/profiles/web/package.json
# "dsh": { "profile": { "bundles": ["@deepseek-ai/dsh-base", "@deepseek-ai/dsh-web-app", "lunheng-article-pipeline"] } }
# 3) 重启 dsh web
安装后技能自动出现在会话的 skill 工具目录,无需手动复制到技能根。
内容
skills/lunheng-article-pipeline/— 技能本体(SKILL.md+AGENTS.md+SOUL.md+references/:7 张角色卡 + 4 个模板 + 运行手册)cordis.patch.yml— bundle 补丁:注册指向包内skills/的 filesystem 技能提供者
按角色分模型(可选)
默认所有角色继承会话模型。若要按角色指派不同模型(检索用便宜快模型,分析/写作/审计用强模型),用随包附带的「分档预设」:
| 工具 | 角色 | 默认模型 | 环境变量 |
|---|---|---|---|
subagent_retrieval |
T1 文献 / T2 数据 / T6 案例 | deepseek-v4-flash |
LUNHENG_RETRIEVAL_MODEL |
subagent_strong |
T3 分析 / T4 写作 | deepseek-v4-pro |
LUNHENG_STRONG_MODEL |
subagent_audit |
T5 审计 | deepseek-v4-pro |
LUNHENG_AUDIT_MODEL |
# 1) 复制预设到用户预设根
cp -r examples/preset "$DSH_HOME/.agent-presets/lunheng"
# 2) 新会话在预设选择器里选「论衡分档」
# 3) 换模型:设环境变量后重启 dsh(模型挂载期求值一次)
export LUNHENG_AUDIT_MODEL=claude-opus-5
dsh web
不装预设也没关系——技能会回退到 subagent,所有角色继承会话模型。详见 examples/preset/README.md 与 docs/installation.md。
验证
dsh --profile web --dump-config # 应能看到 skill-filesystem-lunheng 行
# 新开会话后,skill 工具目录应列出 lunheng-article-pipeline
发布
npm login # 首次需登录
npm publish # 或 npm publish --access public
文档
docs/installation.md— 安装与验证docs/usage.md— 使用流程(五阶段 + 产物结构)docs/architecture.md— 架构(7 角色 + 三角验证 + G0-G11 审计)docs/introduction.md— 插件介绍docs/faq.md— 常见问题CHANGELOG.md— 版本历史CONTRIBUTING.md— 维护与同步指南
许可
MIT,见 LICENSE。
Links
More in this category
GanyuanRan/Aegis★ 1048
Software-engineering method pack for coding agents, with skills for baseline-first planning, systematic debugging, prompt hygiene, verification before completion, and repair/retirement tracking.
superdesigndev/superdesign-skill★ 429
Design skill for UI and marketing graphics on the Superdesign canvas: reads the repo for context, extracts its design system, then generates and iterates branchable design drafts, flow pages, and reusable components through the Superdesign CLI.
linhay/harmony-next.skills★ 328
HarmonyOS NEXT skill bundle for DeepSeek Harness with offline API references and DevEco, HDC, and emulator automation guidance.
Vladimir-Human/humanizer-ru#dsh★ 101
Cleans AI traces from Russian text: finds chatbot copy-paste artefacts (ChatGPT, Gemini, Grok, Perplexity, DeepSeek) and rewrites the text into natural prose on request; 39 regex markers with an evidence registry, offline, text-only bundle.
Vladimir-Human/ru-marketplace-mcp#dsh★ 64
Skills and optional MCP rows for ten Russian marketplaces: price comparison across Wildberries, Detsky Mir and Yandex Market, plus per-source search, product cards and reviews. The 13 skills load on install; both MCP rows stay disabled until RU_MARKETPLACE_MCP_DIR points at a local clone, which needs Python 3.12+ and uv.
dhicoc/dsh-reverse-skill★ 30
Complete reverse-skill pack (85 SKILL.md) as a DeepSeek Harness Cordis plugin: reverse engineering, authorized pentesting and security-research skill router.