Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add TashanGKD/tashan-writing-system --skill document-pipelinegit clone --depth 1 https://github.com/TashanGKD/tashan-writing-systemWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/tashangkd/tashan-writing-system/document-pipeline)<a href="https://agentmods.dev/skills/tashangkd/tashan-writing-system/document-pipeline"><img src="https://agentmods.dev/badge/skills/tashangkd/tashan-writing-system/document-pipeline/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/tashangkd/tashan-writing-system/document-pipeline"><img src="https://agentmods.dev/badge/skills/tashangkd/tashan-writing-system/document-pipeline.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00167 | $0.12286 |
| Opus 5.5 | $0.00067 | $0.04914 |
| Sonnet 5.5 | $0.00033 | $0.02457 |
| Haiku 4.5 | $0.00017 | $0.01229 |
Grade A, and why
document-pipeline scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
98% identical to document-pipeline — 2 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 823 lines — stays where its author put it; the contents beside it link to each section on GitHub.
统一写作流水线(document-pipeline)
适用范围:任何类型的文档——知识库文章、微信公众号、学术论文、技术报告、产品分析等 核心价值:一条完整的、任意阶段可入场的写作闭环,最终产出带原始 Mermaid + qwen-image 嵌入的 MD 手稿,以及可选的格式转换 与旧 Skill 的关系:wechat-article-writer / general-article-writer / research-output 均为本 Skill 的 thin wrapper,触发词不变,内部转发到此处
知识导航表(执行前必须理解的概念根)
| 层级 | 文档 | 需要理解的概念 |
|---|---|---|
| D0 认知根(必读) | _内部总控/认知结构/L1_系统性文档/待建维度/写作习惯与风格手册.md(待建,暂用 _内部总控/AI思维碎片/写作习惯与风格手册.md) |
郑总的写作风格偏好:文章结构/语言风格/禁用套话/标题原则(D0是写作的认知根) |
| D3 规范参考 | .cursor/skills/article-proofreading/SKILL.md(作为审稿标准参考,非触发执行) |
五轮审稿标准:AI腔/标题错误/绝对表达/结构层次/结语完整性 |
| D4 运行时数据 | — | 写作时无固定运行时数据;从任意Stage入场时读取已有草稿 |
核心概念速查: ① 流水线 = 8个Stage,可任意入场;有草稿直接从审稿Stage开始,不必从研究开始 ② 配图是独立Stage:先完成文字内容,再统一做配图,不混合执行 ③ 真实性校对先于审稿:事实正确→逻辑正确→文风优化,顺序不可颠倒
触发判断(先执行)
触发条件:
IF 用户表达了「产出一篇文档/文章/报告」的意图
OR 用户提供了草稿并要求「继续/检查/配图/审稿/转格式」
OR 用户明确说「从 Stage X 开始」/ 「只要做 Y」
THEN 执行 Stage 0,确定入场点和目标格式
不应触发的情况(让 AI 直接回答即可):
- 用户只问一个知识性问题(「XX 是什么?」)
- 用户要修改代码 → role-前端/后端开发
- 用户要更新 L1 认知结构文档 → cognitive-update-knowledge
- 用户要做学术 SI 补充材料 → academic-si-writer(全局 Skill)
Stage 0:入场检测(必须执行)
确认 3 件事:① 从哪个 Stage 开始 ② 目标格式 ③ 若是调研模式,确认维度
A. 判断 target_format:
- 用户说「公众号/微信/WeChat/推文」→ target_format = html
- 用户说「学术论文/LaTeX/论文/期刊」→ target_format = latex
- 用户说「知识库/认知结构/调研笔记/研究报告」→ target_format = md + mode = research
- 其他/未说明 → target_format = md(默认,不询问,直接继续)
**文档性质判断(写入位置决策)**:
- **原创思考类**(用户自己的分析/产品定义/架构设计/认知洞见)→ 写入认知结构 L1/L2 或项目目录
判断依据:内容以用户第一人称视角表达、包含「我的判断」「我认为」或明显是用户原创的产品/架构/认知文档
- **外部调研类**(基于外部来源整理的知识)→ 写入认知结构知识库(REF-EXT)
判断依据:引用了外部文献/官方文档/第三方数据、内容主要是「X工具是什么/X框架如何使用」
- **内部经验积累类**(实际执行中产生的踩坑记录/最佳实践/操作规范)→ 写入总规范库
判断依据:「我们踩过的坑」「部署时发现」「实际操作步骤」「配置参数」等实操内容;
路由规则:确认主题分类 → 写入 `_内部总控/开发规范/[主题]/踩坑速查.md` 或 `最佳实践.md`
主题示例:AI服务器助手/ 、前端规范/、部署规范/、后端规范/ 等;若主题子目录不存在则先创建
⚠️ 若无法判断是「原创思考」还是「内部经验」,询问:「这是你自己的新思考,还是实际执行中总结的经验?」
B. 判断入场 Stage(规则1优先于规则2,触发词优先):
规则1:用户说了明确的操作关键词时,直接采用,不检查草稿信号:
- 用户提供了完整草稿(有目录/多节/超过500字)+ 说「检查逻辑/审核结构/逻辑有没有问题」→ 入场 Stage 2
- 用户提供了完整草稿(有目录/多节/超过500字)+ 说「核实/真实性/事实有没有错/来源」→ 入场 Stage 3
- 用户提供了完整草稿(有目录/多节/超过500字)+ 说「配图/画图/加图/从画图开始」→ 入场 Stage 4
- 用户提供了完整草稿(已有图)+ 说「审稿/给审稿者/交给审核」→ 入场 Stage 7
- 用户提供了已审稿的 MD + 说「转 HTML/转 LaTeX/发布」→ 入场 Stage 8
- 用户说「写一篇/帮我写/我想写/研究一下并写成文档」→ 入场 Stage 1
- **用户说「完善[已有文档]/给[已有文档]加图/补图/按规范完善/这篇文档缺图」+ 明确指向已存在的知识库文档(REF-EXT)→ mode=improve,入场 Stage 2(仅执行图片识别+生成,跳过 Stage 1 研究和 Stage 8D 注册)**
- 无法判断 → 询问「你现在在哪个阶段?有现成草稿吗?」
C. 若 mode = research,判断文档所属维度(存储路径用):
- 技术/工具/架构/系统/API → 系统架构思维维度
- 认知科学/学习/注意力 → 认知科学基础维度
- 产品设计/商业/市场/用户研究 → 产品视野维度
- 写作/内容/表达/传播 → 写作维度
- 研究方法/学术规范/论文写作 → 研究范式维度
- 组织管理/团队/协作/运营 → 组织管理维度
research 模式的 6 个维度选项:
① 系统架构思维维度:技术/工具/架构/系统/API/框架/开源项目
② 认知科学基础维度:认知/学习/注意力/记忆/心理/神经
③ 产品视野维度:产品设计/商业/市场/用户研究
④ 写作维度:写作/内容/表达/传播/媒体
⑤ 研究范式维度:研究方法/学术规范/论文写作
⑥ 组织管理维度:组织/团队/协作/运营/管理
- 多维度命中(话题横跨多个维度)→ 列出候选维度询问用户选择
- 零维度命中(话题不在任何维度关键词中)→ 默认存入①系统架构思维维度,标注 [待分类]
- 完全无法判断 → 列出以上 6 个选项让用户选择
Stage 0 输出:[target_format, entry_stage, dimension(research 模式)]
⚠️ 单任务声明:整条流水线(Stage 0-8)视为一次完整任务。
auto-experience-hook 仅在流水线最终完成(Stage 8 结束或用户明确终止)时触发一次,
不在每个 Stage 完成时独立触发。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 823 lines · 167 tokens per session scan A 55a276ce4da8
document-pipeline is a skill published in the GitHub repository TashanGKD/tashan-writing-system (2 stars, last pushed 7mo ago), licensed MIT. It adds 167 tokens to every session and 12,286 once invoked, about $0.0007 per session on Opus 5.5. A static security scan graded it A with 0 findings. It is 98% identical to document-pipeline, differing in 2 lines, and is treated as a copy.
document-pipeline compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| document-pipelineTashanGKD/tashan-writing-system | Skill | 2 | 7mo ago | MIT | A | 167 | Claude Code, Codex |
| orbit-notionnexu-io/open-design | Skill | 106k | yesterday | Apache-2.0 | A | 117 | Claude Code, Codex |
| instrument-data-to-allotropeanthropics/knowledge-work-plugins | Skill | 35k | 2d ago | Apache-2.0 | A | 123 | Claude Code |
| apify-youtube-transcript-scrapersickn33/agentic-awesome-skills | Skill | 49k | yesterday | MIT | A | 48 | Claude Code |
| legal-holdanthropics/claude-for-legal | Skill | 9.9k | 3d ago | Apache-2.0 | A | 68 | Claude Code |
| ppt-reviewOrkas-AI/Orkas | Skill | 9.0k | yesterday | MIT | A | 2 | Claude Code, Codex |
| feishuHmbown/CodeWhale | Skill | 41k | 5d ago | MIT | A | 33 | Claude Code, Codex |