browser-assisted-hunting
浏览器交互式挖洞:登录态操作、双账号越权对比、SPA渲染取证、DOM XSS确认、验证码登录、前端隐藏功能绕过。 Use when CLI/HTTP evidence is insufficient and the target needs real browser interaction to confirm.
Skill by langbyyi
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/langbyyi-browser-assisted-hunting
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/browser-assisted-hunting
What's inside
- Skill name
browser-assisted-hunting- Description
- 浏览器交互式挖洞:登录态操作、双账号越权对比、SPA渲染取证、DOM XSS确认、验证码登录、前端隐藏功能绕过。 Use when CLI/HTTP evidence is insufficient and the target needs real browser interaction to confirm.
- License
- Apache-2.0
- Instructions
- 53 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
## 定位 命令行与 HTTP 工具拿不到**渲染后状态与交互反馈**。当目标的漏洞需要真浏览器操作才能确认时用本 skill(依赖外部 MCP,如 Playwright)。它只补验证手段,不降低任何证据门槛;授权边界遵循 `authorized-attack-scope`。 ## 何时必须上浏览器 | 场景 | 命令行为什么不行 | 浏览器怎么做 | |---|---|---| | SPA 动态渲染 | 源码无数据,接口运行时才生成 | `browser_snapshot` 读无障碍树,定位真实接口与元素 | | 登录态多步交互 | 登录流有 JS 挑战/跳转/加密 | 表单填写走完登录,再操作目标功能 | | 越权对比 | 需两个独立会话 Cookie 环境 | 双浏览器 context 隔离登录,互相重放 | | DOM XSS | 源码含 sink ≠ 执行 | 渲染后读 DOM/console + 截图执行效果 | | CSRF 实操 | 需浏览器真实携带 Cookie 跨站提交 | 构造表单自动提交,观察副作用 | | 前端隐藏功能 | hidden/disabled 元素命令行看不见 | snapshot 找到元素,绕过前端限制打后端 | ## 标准打法 ### 1. 双账号越权对比(最常用) 1. 开两个浏览器 context(Cookie 隔离),分别登录测试账号 A、B。 2. A 中定位目标功能:订单/个人资料/文件下载/消息等,记下请求 URL 与参数。 3. B 中重放:改 ID 参数直接导航或重新发起,读取渲染结果。 4. B 能看到非 B 的数据即越权成立。证据:B 侧截图 + 快照关键节点 + 命中的 URL/参数。 ### 2. DOM XSS 确认 1. 注入点提交 payload 后等渲染完成。 2. `browser_snapshot` 或执行 JS 读取 sink 处的实际值,确认 payload 进入且被解析执行。 3. 截图执行效果(弹窗/样式变化/外带请求)。「源码里看得到 payload」不算证据,**渲染后执行效果**才算。 ### 3. 登录流与验证码 1. `browser_type` 填表单提交;遇验证码先截图。 2. 截图传给 `ddddocr` 工具识别(`ocr` 文字码 / `detect` 点选坐标 / `slide` 滑块缺口),结果回填。 3. 登录态建立后继续目标操作;登录成功本身可作为「auth_required」字段的实测依据。 ### 4. 前端逻辑绕过 1. snapshot 找 disabled/hidden 的按钮、被前端拦截的接口。 2. 前端限制不代表后端校验:直接调用后端接口或修改元素状态后提交。 3. 后端未校验即成立;证据取请求/响应差分,浏览器只是发现手段。 ## 证据衔接(不可降级) - 浏览器工具调用记录即真实探测证据,`record_vulnerability` 可复现门认可。 - 截图是给人看的补充,**必须**配结构化证据:请求 URL/参数 + 渲染后快照的关键内容。 - 负结果也落 Fact:「浏览器实测 XX 无渲染执行 / 无越权」,防止重复验证。 ## 降级 `tool_search` 查不到 browser_* 工具(未挂浏览器 MCP)时:退回 HTTP 工具 + 手动 Cookie 头模拟登录态;在 Fact 中标注「缺渲染态证据」,不虚构浏览器结果。
From langbyyi/CyberStrikeAI-SRC, Apache-2.0.
More from langbyyi/CyberStrikeAI-SRC
- 0.735score
burp-mcp-vuln-check
langbyyi
Automate low-impact web vulnerability verification through Burp MCP.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
dependency-confusion
langbyyi
Supply-chain testing via package-manager dependency confusion.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
clickjacking
langbyyi
Clickjacking playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
cmdi-command-injection
langbyyi
Command injection playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
csrf-cross-site-request-forgery
langbyyi
CSRF testing playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
crlf-injection
langbyyi
CRLF injection playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
cors-cross-origin-misconfiguration
langbyyi
CORS misconfiguration testing playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
business-logic-vulnerabilities
langbyyi
Business logic vulnerability playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
api-sec
langbyyi
Entry P1 category router for API security.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
authbypass-authentication-flaws
langbyyi
Authentication bypass testing playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
ssti-server-side-template-injection
langbyyi
SSTI playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
deserialization-insecure
langbyyi
Insecure deserialization playbook.
SkillAFreshness: A★ 131Claude Code +9
Signals
- Stars
- 131
- Forks
- 11
- Open issues
- 0
- Last commit (this folder)
- Aug 14, 2026 (2 months ago)
- Last release
- Oct 16, 2026 (11 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- Apache-2.0
- Repository
- github.com/langbyyi/CyberStrikeAI-SRC
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from langbyyi
- 0.735score
burp-mcp-vuln-check
langbyyi
Automate low-impact web vulnerability verification through Burp MCP.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
ssti-server-side-template-injection
langbyyi
SSTI playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
api-sec
langbyyi
Entry P1 category router for API security.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
authbypass-authentication-flaws
langbyyi
Authentication bypass testing playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
business-logic-vulnerabilities
langbyyi
Business logic vulnerability playbook.
SkillAFreshness: A★ 131Claude Code +9 - 0.730score
clickjacking
langbyyi
Clickjacking playbook.
SkillAFreshness: A★ 131Claude Code +9
Cite this page
browser-assisted-hunting skill. Agent Mart. https://agentmart.dev/skills/langbyyi-browser-assisted-hunting (last updated 2026-10-24)