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 agentmods add skills/channinglua/prax-agent/browser-scrapenpx skills add ChanningLua/prax-agent --skill browser-scrapegit clone --depth 1 https://github.com/ChanningLua/prax-agentWhat 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 | $0.00039 | $0.01034 |
| Opus 5 | $0.00019 | $0.00517 |
| Sonnet 5 | $0.00008 | $0.00207 |
| Haiku 4.5 | $0.00004 | $0.00103 |
Grade A, and why
browser-scrape 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 2d 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 — 83 lines — stays where its author put it; the contents beside it link to each section on GitHub.
浏览器抓取技能（AutoCLI）
复用用户本机 Chrome 的登录态抓取需要登录的站点（推特/X、知乎、Bilibili、Reddit、小红书等 55+ 平台）。不需要 cookie 配置、不需要 API key。
前置要求（由用户完成一次即可）
- 安装 AutoCLI Rust 二进制：参见 https://github.com/nashsu/AutoCLI（单文件 ~4.7MB，无运行时依赖）
- 装 autocli Chrome 扩展（仓库 README 里有下载链接和加载步骤）
- 保持 Chrome 运行并在目标站点处于登录态
autocli doctor应报告OK
如果 autocli doctor 失败，先提醒用户修复前置条件，不要继续抓取。
能力范围
AutoCLI 在 Prax 里通过普通 Bash 工具调用——它不是 MCP server，只是一个 CLI。只要 PATH 里能找到 autocli，任何 Prax agent 都能用。
常用命令（按频次排序）
# 诊断（第一次使用必跑）
autocli doctor
# 推特/X
autocli twitter timeline --limit 20 --format json
autocli twitter search --query "AI safety" --limit 10 --format json
# 知乎
autocli zhihu hot --limit 20 --format json
# Bilibili
autocli bilibili hot --limit 10 --format json
# Reddit
autocli reddit subreddit --name "MachineLearning" --limit 15 --format json
# 任意网页文章（不走登录态）
autocli read https://example.com/article --format md
autocli read https://example.com/article --format text -o /tmp/article.txt
--format json|md|text|yaml|csv任选；编程任务优先json，归档任务用md--limit N控制条数autocli --help查所有子命令；单平台用autocli twitter --help
典型抓取流程
用户让我"抓今天 X 上 AI 相关点赞 top 10 存到 Obsidian"，我的动作：
autocli doctor确认前置条件autocli twitter timeline --limit 50 --format json拉最近推文- 本地过滤（按关键词/点赞数），用 Write 工具存到
.prax/vault/ai-news-hub/YYYY-MM-DD/下 - 每条一个 markdown 文件，header 带
tweet_id / author / likes / url / scraped_at - 完成后总结文件路径给用户
产出约定（配合 knowledge-compile 技能）
- 目录命名：
.prax/vault/<topic>/YYYY-MM-DD/（方便 Obsidian 按日期归档） - 文件命名：
<source>-<id>.md，例如twitter-17xxx.md - 文件 frontmatter：至少包含
source / url / scraped_at，供下游knowledge-compile编译 wiki 用 - 原始 JSON 保留：把
autocli ... --format json的原始输出同步存一份到raw/<source>-<stamp>.json，便于溯源
边界与禁区
- 不要用 AutoCLI 做 写操作（发帖/点赞/关注）除非用户明确要求；无脑批量操作会导致账号封禁
- 不要把登录 cookie 或扩展私钥外传
- 被站点频控或返回异常时先停下来报告，不要循环重试——可能会触发风控
配合其他技能
- 抓完 →
knowledge-compile把散文件整理成 wiki - 每天定时运行 →
cron里配prax cron add用本技能 - 出成品 →
Notify工具把结果推到飞书/邮箱
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.
- 2d ago First seen · 83 lines · 39 tokens per session scan A 42866d05a5f5
browser-scrape is a skill published in the GitHub repository ChanningLua/prax-agent (272 stars, last pushed 1mo ago), licensed MIT. It adds 39 tokens to every session and 1,034 once invoked, about $0.0002 per session on Opus 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 2026-08-30.
Other skills, from other repositories
scraper-brief
Writes a clear, technical brief describing exactly what data needs to be collected from a website or set of web pages, how it should be structured, and what edge cases and legal/ethical considerations apply — for handoff to a developer or data team.
scrapling-official
Scrape web pages using Scrapling with anti-bot bypass (like Cloudflare Turnstile), stealth headless browsing, spiders framework, adaptive scraping, and JavaScript rendering. Use when asked to scrape, crawl, or extract data from websites; webfetch fails; the site has anti-bot protections; write Python code to…
browser-automation
Playwright-based browser automation patterns for autonomous web interaction.
browser-automation
Playwright-based browser automation patterns for autonomous web interaction.
web-browse
Drive a headless browser to navigate pages, read content, click, and fill forms — for sites that need JavaScript rendering or interaction beyond a plain HTTP fetch.
ecommerce-full-pipeline
电商运营在开展跨境电商或闲鱼捡漏业务时，若需解决选品难、上架繁琐等痛点，必用此技能！一键打通“爆品挖掘→1688采集→多平台上架→推广文案→短视频生成”全自动流水线，轻松实现端到端自动化，让开店运营效率翻倍。.