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 shawnwei512/doc-distillation-mcp --skill skillgit clone --depth 1 https://github.com/shawnwei512/doc-distillation-mcpWrote 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/shawnwei512/doc-distillation-mcp/skill)<a href="https://agentmods.dev/skills/shawnwei512/doc-distillation-mcp/skill"><img src="https://agentmods.dev/badge/skills/shawnwei512/doc-distillation-mcp/skill/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/shawnwei512/doc-distillation-mcp/skill"><img src="https://agentmods.dev/badge/skills/shawnwei512/doc-distillation-mcp/skill.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 | $0.00172 | $0.10894 |
| Opus 5.5 | $0.00069 | $0.04358 |
| Sonnet 5.5 | $0.00034 | $0.02179 |
| Haiku 4.5 | $0.00017 | $0.01089 |
Grade A, and why
doc-distillation 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.
How it starts
The opening of the file, as written. The whole thing — 673 lines — stays where its author put it; the contents beside it link to each section on GitHub.
资料蒸馏技能 (Document Distillation)
将各类来源（飞书文档、网页、PDF等）的原始资料，通过整理、分析、结构化、重组，生成两种高质量输出：HTML蒸馏文稿 和 Obsidian笔记。
合规使用边界
[!important] 本技能仅用于个人学习目的的知识整理
适用场景（用户需确保对内容拥有合法访问权）：
- 用户自己创建的飞书文档/笔记
- 用户被授权访问的团队文档
- 公开免费的网页/视频/播客内容
- 用户已购买的课程（仅限个人学习笔记，不公开发布）
不适用场景：
- 未经授权访问付费/受限内容
- 批量复制他人版权内容用于分发
- 绕过任何平台的技术保护措施
蒸馏原则：蒸馏是知识提炼（用自己的话概括核心观点），不是原文复制。输出内容应大幅压缩原文，保留信息密度而非原文照搬。
触发条件
当用户出现以下意图时激活：
- 提供文档链接（飞书/网页等）并要求"提取"、"蒸馏"、"整理"、"总结"
- 提供视频链接（YouTube/B站/抖音/快手/小红书/视频号等）并要求"提取文案"、"蒸馏"、"整理"
- 提供播客链接或本地音视频文件并要求"转录"、"蒸馏"
- 要求将文档/视频内容转化为笔记或文稿
- 提及"蒸馏"一词与文档/视频处理相关
- 要求从资料中提炼核心知识和结构
核心原则
- 双输出：始终生成 HTML蒸馏文稿 + Obsidian笔记，缺一不可
- 原文尊重：蒸馏是提炼而非改写，保留原文核心观点和表达
- 结构优先：先理解整体结构，再填充细节
- 图片完整：提取所有图片，保持图文关联
- 图片内容蒸馏（关键！）：不仅要提取图片本身，还必须对包含文字信息的图表、框架图、流程图等做内容蒸馏——用文字完整提取图片中的所有文字、结构和关系，融入Obsidian笔记中。这是确保笔记"可独立阅读"的核心要求。
- 独立成篇：每篇文档独立生成一套输出，不合并多篇
- 图片智能过滤（效率优先）：不盲目下载所有图片，通过三层过滤机制（自动规则→上下文预判→选择性蒸馏），只下载和蒸馏含关键信息的图片，节约token和时间
- 完整性优先于效率（关键！）：节约token的前提是不遗漏关键信息。压缩表达方式，不压缩信息量。通过"结构骨架法+关键要素标记+双向校验"三重机制，在蒸馏前建立完整性清单，在蒸馏后执行缺口检查，确保零遗漏。
token节约与完整性的平衡原则
[!important] 核心准则：压缩表达方式，不压缩信息量 蒸馏的目标是用更少的字数传达同样的信息密度，而非用更少的字数传达更少的信息。
| 区域 | 策略 | 说明 |
|---|---|---|
| 高信息密度区（不可压缩） | 原样保留或结构化重组 | 核心论点、关键数据（数字/比例/金额）、操作步骤、公式/模板、框架/模型、清单/表格、避坑提醒 |
| 低信息密度区（可压缩） | 提炼为一句话或删除 | 过渡性描述、重复论述、冗长案例、情感修辞、背景铺垫 |
| 图片过滤区 | 三层过滤+安全网 | 装饰图跳过，内容图必须蒸馏，边界图存疑时保留 |
用户偏好（务必遵守）
- 沟通语言：中文
- HTML蒸馏文稿路径：
~/Documents/蒸馏文稿/ - Obsidian笔记路径：
~/Documents/obsidian/（具体子目录由用户每次指定） - 图片存储：Obsidian笔记目录下的
assets/子目录 - 不需要：小红书帖子等其他格式输出
蒸馏工作流程
阶段1：来源识别与内容提取
1.1 识别来源类型
- 飞书文档（feishu.cn / larkoffice.com / yitang.top/fs-doc）：优先尝试 lark-doc 技能，权限不足时用浏览器提取
- 普通网页：直接用浏览器提取
- PDF/PPT：用 pdfplumber/PyMuPDF 提取文本，PPT型PDF需渲染页面为图片后Read蒸馏
- 视频/播客（YouTube/B站/抖音/快手/小红书/视频号/播客）：用
scripts/video_transcript.py提取文案，详见references/video-transcript-guide.md - 本地音视频文件（mp3/m4a/mp4/wav等）：直接用
scripts/video_transcript.py转录 - 用户粘贴文本：直接进入阶段2
What ships with it
9 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- examples/视频文案蒸馏示例.md 799 B
- examples/飞书文档蒸馏示例.md 930 B
- references/feishu-image-extraction-guide.md 8.4 KB
- references/html-template.md 7.7 KB
- references/obsidian-template.md 5.1 KB
- references/quality-checklist.md 8.9 KB
- references/video-transcript-guide.md 8.0 KB
- scripts/image_preprocessor.py 6.6 KB runs code
- scripts/video_transcript.py 16 KB runs code
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 Changed · +34 lines · +14 tokens per session c1fe3eebb6ac
- 2mo ago First seen · 639 lines · 158 tokens per session scan A 214aeffdea36
doc-distillation is a skill published in the GitHub repository shawnwei512/doc-distillation-mcp (1 stars, last pushed 2mo ago), licensed MIT. It adds 172 tokens to every session and 10,894 once invoked, about $0.0007 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-04.
doc-distillation compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| doc-distillationshawnwei512/doc-distillation-mcp | Skill | 1 | 2mo ago | MIT | A | 172 | Claude Code, Codex |
| orbit-notionnexu-io/open-design | Skill | 106k | today | Apache-2.0 | A | 117 | Claude Code, Codex |
| instrument-data-to-allotropeanthropics/knowledge-work-plugins | Skill | 35k | today | Apache-2.0 | A | 123 | Claude Code |
| apify-youtube-transcript-scrapersickn33/agentic-awesome-skills | Skill | 49k | today | MIT | A | 48 | Claude Code |
| legal-holdanthropics/claude-for-legal | Skill | 9.9k | today | Apache-2.0 | A | 68 | Claude Code |
| ppt-reviewOrkas-AI/Orkas | Skill | 9.1k | today | MIT | A | 2 | Claude Code, Codex |
| feishuHmbown/CodeWhale | Skill | 41k | 6d ago | MIT | A | 33 | Claude Code, Codex |