Token导航 LogoToken导航

Agent Skill · 研究检索

filter-js-from-html(filter JS from HTML 搜索)

filter-js-from-html 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

filter-js-from-html 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为操作浏览器;许可证为 MIT,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 45,550。

总安装 840周安装 35GitHub Stars 93下载 280核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:filter-js-from-html(filter JS from HTML 搜索)
来源仓库:https://github.com/letta-ai/skills
仓库路径:skills/filter-js-from-html
安装命令:
npx skills add https://github.com/letta-ai/skills --skill filter-js-from-html
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/letta-ai/skills --skill filter-js-from-html

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

filter-js-from-html 用于查找、检索和筛选相关信息。适合根据关键词快速定位候选结果,支持多种宿主环境。通过 npx skills add 命令安装,需确认权限范围和维护状态。使用前应核实是否会触发联网、命令执行或文件读写操作。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/filter-js-from-html

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/letta-ai/skills --skill filter-js-from-html;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要操作浏览器,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

filter-js-from-html 怎么安装?

在终端执行 npx skills add https://github.com/letta-ai/skills --skill filter-js-from-html。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

filter-js-from-html 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

filter-js-from-html 需要什么权限?

权限级别是操作浏览器。该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

filter-js-from-html 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

filter-js-from-html 用的人多吗?

公开数据是累计安装 840 次,近一周安装 35 次,下载 280 次,GitHub 93 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

filter-js-from-html 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
filter-js-from-html

filter JS from HTML 搜索 · 本页

84093操作浏览器MIT通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Filter JavaScript from HTML

Overview

This skill provides guidance for tasks that require removing JavaScript and XSS attack vectors from HTML content while preserving the original formatting exactly. The key challenge is balancing comprehensive security filtering with format preservation.

Critical Requirements Analysis

Before implementation, identify and prioritize these requirements:

  1. Security completeness: All XSS vectors must be removed
  2. Format preservation: Output must be functionally identical to input except for harmful content removal
  3. Clean content handling: Files without XSS content should remain completely unchanged

These requirements often conflict - comprehensive parsing may alter formatting, while simple string replacement may miss attack vectors.

Approach Selection

Option 1: Regex-Based Surgical Removal (Recommended for Format Preservation)

When the task explicitly requires preserving original formatting, prefer regex-based approaches that surgically remove only the dangerous content.

Advantages:

  • Preserves whitespace, attribute ordering, quote styles exactly
  • Does not reconstruct or reformat HTML
  • Output matches input character-for-character except for removed content

Considerations:

  • Requires careful pattern construction to avoid partial matches
  • Must handle various encodings and obfuscation techniques
  • Test patterns against comprehensive XSS vector lists

Option 2: HTML Parser-Based Filtering

When format preservation is less critical or when dealing with malformed HTML.

Considerations:

  • HTML parsers inherently reconstruct output, changing formatting
  • May normalize attribute quotes, whitespace, tag casing
  • Better for malformed HTML that regex cannot reliably parse
  • If using this approach, verify that clean HTML files remain unchanged

Comprehensive XSS Vector Checklist

Before implementing, research and account for ALL of these attack categories:

1. Script Execution Tags

  • <script> tags (including variations with attributes)
  • <noscript> abuse cases

2. Event Handlers (Comprehensive List Required)

Common handlers:

  • onclick, onload, onerror, onmouseover, onfocus, onblur

Frequently missed handlers:

  • onlayoutcomplete, ontimeerror, onselectionchange
  • onrowsinserted, onrowsdelete, onrowexit, onrowenter
  • oncellchange, ondataavailable, ondatasetchanged, ondatasetcomplete
  • onbeforeupdate, onafterupdate, onerrorupdate
  • onfilterchange, onpropertychange, onreadystatechange
  • onbeforeprint, onafterprint, onbeforeunload
  • oncontextmenu, ondrag, ondragend, ondragenter, ondragleave
  • ondragover, ondragstart, ondrop
  • onhashchange, oninput, oninvalid, onpageshow, onpagehide
  • onpopstate, onresize, onstorage, onwheel

Action: Search for comprehensive event handler lists (e.g., MDN, OWASP) rather than relying on memory.

3. JavaScript URL Protocol

  • javascript: in href, src, action, formaction, data, poster attributes
  • Case variations: JavaScript:, JAVASCRIPT:, JaVaScRiPt:
  • Encoded variations: javascript:, javascript:

4. Other Dangerous Protocols

  • vbscript: (IE legacy)
  • data: URIs with script content: data:text/html,<script>...</script>
  • data:text/html;base64,... encoded payloads

5. CSS-Based Attacks

  • <style> tags with dangerous properties
  • -moz-binding (Firefox legacy)
  • expression() (IE legacy)
  • behavior: property
  • @import with javascript or data URIs

6. Meta Tag Attacks

  • <meta http-equiv="refresh" content="0;url=data:text/html,...">
  • <meta http-equiv="refresh" content="0;url=javascript:...">

7. External Resource Loading

  • <link> tags with dangerous href values
  • <object> tags with data attributes
  • <embed> tags with src attributes
  • <applet> tags (legacy)
  • <iframe> with src or srcdoc containing scripts

8. SVG-Based Attacks

  • <svg onload="..."> and other SVG event handlers
  • <svg><script>...</script></svg>
  • SVG <use> with external references

9. Encoding and Obfuscation

  • HTML entity encoding: <script>
  • URL encoding: %3Cscript%3E
  • UTF-7 encoding attacks
  • Null byte injection: <scr\0ipt>
  • Unicode variations

10. HTML Comment Exploits

  • Conditional comments: <!--[if IE]><script>...<![endif]-->
  • Nested comment breaking

Verification Strategy

Test Categories (All Required)

  1. XSS Attack Vectors

- Use established XSS test suites (OWASP XSS Filter Evasion Cheat Sheet) - Test XSS polyglots that combine multiple techniques - Include lesser-known event handlers in tests

  1. Format Preservation

- Provide clean HTML files with varied formatting - Verify byte-for-byte identical output for clean files - Test various whitespace patterns, quote styles, attribute ordering

  1. Edge Cases

- Malformed HTML - Mixed case tags and attributes - Attributes without quotes - Multiple encodings in same document

Testing Process

  1. Research first: Before writing tests, search for:

- OWASP XSS Prevention Cheat Sheet - XSS Filter Evasion Cheat Sheet - Known XSS polyglots - Browser-specific attack vectors

  1. Create adversarial tests: Do not rely solely on self-created test cases

- Use external comprehensive test suites - Include vectors that have bypassed filters historically

  1. Test clean content preservation: Equal priority to security testing

- Create diverse clean HTML samples - Verify no modifications occur - Check whitespace, comments, attribute order

Common Pitfalls

1. Incomplete Event Handler Lists

Mistake: Hardcoding only common event handlers like onclick, onload, onerror. Solution: Research and include ALL valid HTML event handlers, including deprecated and browser-specific ones.

2. Ignoring CSS Attack Vectors

Mistake: Focusing only on JavaScript while ignoring CSS-based XSS. Solution: Filter <style> tags, dangerous CSS properties, and style attributes with expressions.

3. Missing Protocol Handlers

Mistake: Only filtering javascript: protocol. Solution: Also filter vbscript:, data: URIs with dangerous content, and handle encoded protocol names.

4. Format Alteration with Parsers

Mistake: Using HTML parsers when format preservation is required. Solution: If format preservation is critical, use regex-based surgical removal or verify parser output matches input formatting.

5. Self-Validating Tests

Mistake: Creating test cases that match implementation capabilities rather than real attack vectors. Solution: Use external, adversarial test suites created by security researchers.

6. Quote and Encoding Handling

Mistake: Not handling HTML entities in attributes (", '). Solution: Consider how encoded characters in attributes might bypass filters.

7. Forgetting Meta Refresh

Mistake: Not filtering <meta http-equiv="refresh"> with dangerous URLs. Solution: Include meta tags in the filtering scope, especially those with data: or javascript: URLs.

8. Ignoring External Resources

Mistake: Not filtering <link>, <object>, <embed> tags. Solution: Evaluate whether these tags can load or execute dangerous content.

Implementation Checklist

Before considering the implementation complete:

  • Researched comprehensive XSS attack vector lists
  • Implemented filtering for ALL event handlers (not just common ones)
  • Handled script tags and noscript abuse
  • Filtered javascript:, vbscript:, and dangerous data: URIs
  • Addressed CSS-based attacks (style tags, expressions, bindings)
  • Handled meta refresh attacks
  • Considered link, object, embed, applet tags
  • Handled SVG-based attacks
  • Accounted for encoding variations
  • Tested with external XSS test suites
  • Verified clean HTML files remain unchanged
  • Tested format preservation (whitespace, quotes, ordering)

平台分布

Claude Code

29.34%
按下载量换算82

Gemini CLI

24.68%
按下载量换算69

Codex

19.37%
按下载量换算54

Antigravity

13.51%
按下载量换算38

OpenCode

8.28%
按下载量换算23

windsurf

3.68%
按下载量换算10

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。

继续浏览同类 Skills