Obsidian Knowledge Pipeline

将URL文章自动抓取、分类、配图、编译入库Obsidian知识库的7步流水线技能，供Agent复用。

Security Scan

The skill's instructions, templates, and optional environment usage are coherent with its stated purpose of fetching URLs and producing Obsidian wiki nodes; it mainly reads/writes your Vault and may call image-generation tools if you provide an API key.

Purpose & Capability

Name/description (URL → Obsidian nodes) match the actual instructions: fetch content, store raw, generate images, classify, compile wiki nodes, and create backlinks. Optional DASHSCOPE_API_KEY for an image provider is consistent with the image-generation step.

Instruction Scope

Instructions legitimately operate on the Obsidian Vault (create raw/, images/, wiki/, Clippings/, update and append files, scan existing .md files to add backlinks). This is expected for the purpose but is high-impact: the skill will read and modify many files in your vault and will append a persistent mapping file containing source URLs and generated prompts (privacy-sensitive). It also references local helper tools/scripts (web_fetch, image_generate, ~/.openclaw/..., ~/.agents/...), which the agent will attempt to call if present.

Install Mechanism

Instruction-only skill with no install spec or downloaded code. That's low-risk. Note: it assumes the presence of other tools/skills (web_fetch, image_generate, scrapling script, Wan2.7 script) that are not installed by this skill; missing dependencies will cause failures but are not malicious.

Credentials

No required credentials declared in registry metadata. SKILL.md documents two environment variables: OBSIDIAN_VAULT_PATH (to locate the Vault) and an optional DASHSCOPE_API_KEY for Wan2.7 image generation. Requesting an API key for image generation is proportionate. Be aware the mapping file stores full prompts and source URLs (may contain sensitive info) and image generation may result in outbound network calls when the API key is supplied.

Persistence & Privilege

Skill is not always-enabled and does not modify other skills. It writes files into the user-specified Vault and may modify many .md files (adding backlinks); this is within the stated scope but is a privileged file-write capability the user should be comfortable granting. Autonomous invocation is allowed by default (platform normal) but not set to always:true.

Assessment

This skill appears to do what it says: it will fetch web articles and write raw files, generated images, a persistent mapping file, compiled wiki nodes, and backlink edits inside the specified Obsidian Vault. Before installing or running: 1) Back up your Vault (the skill will modify many .md files and append a mapping file). 2) Review or sandbox the Vault (use a disposable test vault) to confirm behavior. 3) Only supply DASHSCOPE_API_KEY if you trust the image provider and accept outbound image-generation network calls; prompts and source URLs are stored locally in plain text. 4) Note the skill expects local helper tools/scripts (web_fetch, image_generate, scrapling, Wan2.7) — if those call external services, review them separately. 5) Prefer invoking the skill manually (not granting broad autonomous execution) until you confirm it behaves as expected.

概述

本技能实现从URL到Obsidian知识库节点的全自动处理流水线。任意Agent加载本技能后，即可按照标准化7步流程将网络文章转化为结构化的知识库节点，包含配图、分类、双向链接和知识图谱关系。

环境变量

变量	默认值	说明
OBSIDIAN_VAULT_PATH	~/Documents/Obsidian Vault	Obsidian Vault 根目录
DASHSCOPE_API_KEY	(无默认值)	Wan2.7 配图生成所需 API Key（可选，有则用Wan2.7脚本，无则用 image_generate 工具）

快速参考

# 标准用法（Agent加载本skill后执行）
# 输入：一个或多个URL
# 输出：raw文章 + 配图 + 映射关系 + Wiki编译节点 + 知识图谱链接

# 关键路径
VAULT="${OBSIDIAN_VAULT_PATH:-$HOME/Documents/Obsidian Vault}"
RAW_DIR="$VAULT/raw"
IMAGES_DIR="$RAW_DIR/images"
MAPPING_FILE="$IMAGES_DIR/关联关系.md"
WIKI_DIR="$VAULT/wiki/知识透视"
CLIPPINGS_DIR="$VAULT/Clippings"

7步流程总览

URL → ① 抓取 → ② 存raw/ → ③ 生成配图 → ④ 更新映射 → ⑤ 自动分类 → ⑥ Wiki编译 → ⑦ 知识图谱

步骤	说明	详细文档
① 抓取内容	web_fetch 优先，scrapling 回退	pipeline.md
② 存入raw/	标准模板存储原文	pipeline.md
③ 生成配图	image_generate / Wan2.7	image-mapping.md
④ 更新映射	关联关系.md 标准条目	image-mapping.md
⑤ 自动分类	6大分类规则	classification.md
⑥ Wiki编译	6要素结构化节点	classification.md
⑦ 知识图谱	双向链接建立	pipeline.md

详细文档索引

• pipeline.md — 完整7步处理流程详细说明
• classification.md — 自动分类规则和Wiki编译标准
• image-mapping.md — 配图生成与映射规则

模板文件

• templates/raw-article.md — Raw 文章存储模板
• templates/wiki-node.md — Wiki 节点编译模板
• templates/mapping-entry.md — 映射关系条目模板

禁止事项

• ❌ 禁止使用 assets/ 目录存放配图，统一用 raw/images/
• ❌ 禁止硬编码 Vault 路径，必须通过 $OBSIDIAN_VAULT_PATH 环境变量
• ❌ 禁止硬编码 API Key，通过 $DASHSCOPE_API_KEY 环境变量引用
• ❌ 禁止跳过任何步骤，7步必须顺序执行
• ❌ 禁止在 Wiki 节点中省略6要素中的任何一个
• ❌ 禁止覆盖已有同名文件，遇到冲突时追加时间戳后缀
• ❌ 禁止在映射关系文件中使用非标准格式
• ❌ 禁止忽略 Clippings/ 目录的同步

Clippings/ 同步流程

每次处理完文章后，需将编译结果同步到 Clippings/ 目录：

1. 在 $CLIPPINGS_DIR 下创建与 raw/ 同名的子目录结构
2. 将 Wiki 编译节点软链接或复制到对应位置
3. 确保 Clippings/ 中的 front matter 包含 source_url 字段

使用示例

## Agent 调用流程

1. 读取本 SKILL.md，理解整体流程
2. 获取目标 URL
3. 按 pipeline.md 逐步执行
4. 使用 classification.md 进行分类和 Wiki 编译
5. 使用 image-mapping.md 处理配图
6. 各步骤使用 templates/ 中的标准模板