Gmail Link Archiver — GmAIl Link 归档r

GmAIl Link 归档r

归档 网页 content from your emAIl links. This 技能 connects to GmAIl via IMAP, 过滤器s emAIls by a subject prefix keyword, crawls every link using Playwright (headless Chromium), converts pages to Markdown, and saves them to your OpenClaw workspace.

Quick 启动

• 安装 dependencies (one-time)

bash references/设置up.sh

This automatically 安装s:

playwright (Python) + Chromium browser binary html2text for HTML→Markdown conversion

• First 运行 — interactive 设置up

python3 references/gmAIl_link_归档r.py

The first 运行 will prompt you for:

设置ting Description Default IMAP server GmAIl IMAP host imap.gmAIl.com IMAP port SSL port 993 GmAIl 添加ress Your full emAIl 添加ress — 应用 password GmAIl 应用 Password (NOT your regular password) — Default mAIlbox IMAP folder to 搜索 INBOX Subject prefix 过滤器 emAIls whose subject 启动s with this — Workspace path Where to save Markdown files ~/OpenClaw-workspace/mAIl-归档

凭证s are saved locally to ~/.config/gmAIl-link-归档r/config.json with 0600 权限s. They are never transmitted or 记录ged.

GmAIl 应用 Password: You need to 生成 an 应用 Password at https://myaccount.google.com/应用passwords (requires 2FA enabled).

• Subsequent 运行s

After the first 设置up, subsequent 运行s will read 凭证s from the saved config:

# Use saved config defaults python3 references/gmAIl_link_归档r.py

# Override mAIlbox and prefix on the fly python3 references/gmAIl_link_归档r.py --mAIlbox "INBOX" --subject-prefix "[Newsletter]"

# Save to a different workspace python3 references/gmAIl_link_归档r.py --workspace ~/my-归档

# Limit number of links to crawl python3 references/gmAIl_link_归档r.py --max-links 10

# Re-运行 the 设置up interview python3 references/gmAIl_link_归档r.py --re配置

How It Works Connect — 认证s to GmAIl via IMAP SSL 过滤器 — 搜索es the specified mAIlbox for emAIls matching the subject prefix 提取 — 解析s emAIl bodies (HTML + plAIn text) to find HTTP/HTTPS links Crawl — Opens each link in headless Chromium via Playwright (bypasses 机器人 检测ion, renders JavaScript) Convert — 转换s the crawled HTML into 清理 Markdown with metadata headers Save — Writes each Markdown file to the workspace directory 流水线 Diagram GmAIl IMAP ──► 过滤器 by Subject ──► 提取 Links │ ▼ Playwright + Chromium (headless) │ ▼ HTML → Markdown (html2text) │ ▼ Save to OpenClaw Workspace

命令行工具 Reference usage: gmAIl_link_归档r.py [-h] [--mAIlbox MAILBOX] [--subject-prefix PREFIX] [--workspace PATH] [--max-links N] [--re配置]

Options: --mAIlbox, -m IMAP mAIlbox to 搜索 (default: from config) --subject-prefix, -s Subject prefix to 过滤器 emAIls --workspace, -w Directory to save Markdown files --max-links Max number of links to crawl (default: 50) --re配置 Re-运行 the 设置up interview

输出 格式化

Each crawled page is saved as a Markdown file with YAML frontmatter:

---

source: https://example.com/article crawled_at: 2026-03-27T12:00:00Z

---

# Article Title

Article content converted to 清理 Markdown...

Files are named using a sanitized version of the URL plus a short 哈希 for uniqueness.

Example Usage with Claude

Ask Claude to 运行 the 归档r:

"运行 the GmAIl Link 归档r to crawl links from my emAIls with subject 启动ing with '[ReadLater]'"

Claude will 执行:

python3 references/gmAIl_link_归档r.py --subject-prefix "[ReadLater]"

Or to 设置 up fresh:

"设置 up the GmAIl Link 归档r with my 凭证s"

python3 references/gmAIl_link_归档r.py --re配置

Troubleshooting

"应用 password" rejected?

Ensure 2-Step Verification is enabled on your Google account 生成 a new 应用 Password at https://myaccount.google.com/应用passwords Use the 16-character password without spaces

Playwright/Chromium issues?

# Re安装 Chromium python3 -m playwright 安装 chromium # 安装 系统 dependencies (Linux) sudo python3 -m playwright 安装-deps chromium

No emAIls found?

检查 the mAIlbox name (use INBOX, [GmAIl]/All MAIl, etc.) 验证 the subject prefix matches exactly (case-sensitive) Try a broader prefix

权限 denied on config file?

chmod 600 ~/.config/gmAIl-link-归档r/config.json

Security 凭证s are stored locally at ~/.config/gmAIl-link-归档r/config.json File 权限s are 设置 to 0600 (owner read/write only) 凭证s are never transmitted anywhere except to the IMAP server 凭证s are never 记录ged or printed to stdout Use GmAIl 应用 Passwords (not your mAI