Pilot Browse MCP

中文

MCP toolset + website capability learner for the browser. Agent autonomously explores websites and builds understanding manuals.

---

Showcase (Pi Demo)

Capability Learning

Install a skill. As the agent interacts with a website, it automatically discovers APIs and binds them with browser workflows into reusable capabilities.

Learning output is saved in website-manuals/:

website-manuals/<site>/
  README.md              # Root index
  pages/                 # Page interaction models
  navigation/            # Navigation paths
  workflows/
    README.md            # Workflow index
    flows/               # Workflow JSON files
    scripts/             # PAB automation scripts (.pab files)
  apis/
    README.md            # API index (browse first)
    endpoints/           # API JSON files

Guided Teaching

(Using Bilibili as demo site)

Stuck on complex interactions? Record or mark elements manually to help the agent learn.

Mark Element

Record Workflow

Manual Reuse

Based on the exploration manual, the agent can quickly understand websites and complete tasks faster.

Two no-LLM modes are also supported:

• .pab scripts for automation tasks -- capabilities match the agent's MCP tools
• Python/other scripts for specialized data collection tasks

Demos

Build various workflows, simple demo showcase.

Automation Operations

Agent: Qidian Novel Saver

Search a novel, save the first 5 chapters.

---

.pab Script: Bilibili Open 3 Trending Videos and Comment

---

Python Script Generation

Generate various scripts based on the exploration manual.

For example, a novel scraper script:

I used Pilot to analyze the novel website structure, chapter API, generated scraping scripts, built Docker images, deployed 9 containers on NAS, and crawled over 4000 novels in 2 days.

---

Install

# Build
cd server && npm install && npm run build
cd extension && npm install && npm run build

# Load extension
# chrome://extensions/ -> Developer mode -> Load unpacked -> extension/dist/

# Start server
cd server && node dist/index.js

MCP Configuration

Example - fill args with the actual path to server/dist/index.js:

{
  "mcpServers": {
    "browser-mcp": {
      "command": "node",
      "args": ["/path/to/server/dist/index.js"]
    }
  }
}

Skills

Check scripts/Skill for available skills.

Quick Start with Examples

Refer to agent-examples/ for ready-to-use agent workspace examples. Run your agent in one of those directories and it will automatically load the MCP config, skills, and project prompts. (Remember to update the MCP path to your actual setup.)

---

Architecture

AI Agent (Claude Code / Pi / Codex)
    |
    | 1) MCP stdio protocol (JSON-RPC)
    | stdin / stdout
    v
MCP Server (Node.js)          protocol translator
    |
    | 2) WebSocket :9456
    v
Chrome Extension
    |
    | 3) Chrome API
    |
    v
Browser

Features

• 62 MCP tools -- tab management, content extraction, DOM operations, network interception, file saving, workflow recording, script automation, SQL injection detection, JS reverse
• PAB scripting -- Python-like DSL for browser automation with if/for/fn support. Run from popup, no LLM needed
• Element picker -- click any element on the page and tell the agent what it is
• Workflow recording -- demonstrate operations to the agent, it learns and reuses
• Network API toolkit -- monitor, search, inspect, replay with overrides, export code, and analyze site API structure
• Token-efficient saving -- save page content directly to disk, bypassing the LLM
• Shadow DOM + contenteditable

Tools

Category	Tool	What it does
Page	browser_get_markdown	Convert page to clean Markdown via Readability + Turndown
	browser_get_text	Get plain text of the page (lighter than get_html)
	browser_get_html	Get raw HTML of the page (heavy, last resort)
	browser_find	Find element by visible text, aria-label, or role
	browser_current_page	Get current tab URL and title
	browser_inspect_page	See page structure (headings, sections, buttons)
	browser_query	Query elements by CSS selector (penetrates Shadow DOM)
	browser_evaluate	Execute JS in page context
	browser_extract_article	Extract article metadata (title, author, date, body)
	browser_extract_table	Extract HTML table as JSON array
	browser_extract_links	Extract all links from the page
	browser_extract_images	Extract image info (src, alt, size)
Actions	browser_click	Click an element (composed:true for Shadow DOM)
	browser_type	Type text into input or contenteditable
	browser_scroll	Scroll the page
	browser_wait	Wait for a given number of milliseconds
	browser_wait_for_element	Wait for an element to appear
Saving	browser_save_content	Auto-detect main content and save to file (zero LLM tokens)
	browser_save_xpath	Extract by XPath and save to file
Network	browser_start_network_monitor	Start intercepting requests
	browser_stop_network_monitor	Stop monitoring (cache preserved for replay)
	browser_network_clear_cache	Clear cached requests without stopping monitoring
	browser_network_search	Search cached requests by keyword, method, status
	browser_network_detail	Get full details of a cached request (headers, body, timing)
	browser_network_wait	Wait for a matching request after an action (replaces fixed delay)
	browser_network_replay	Replay with overrides (query/headers/body) + extract JSON path
	browser_network_export	Export request as curl / fetch / Python / HAR
	browser_network_analyze	Analyze API structure of a site from cached requests
	browser_network_override	Set response override rules (body, status, headers)
Tabs	browser_list_tabs	List all open tabs
	browser_open / close / activate	Tab management
Recording	workflow_list_recordings	View recordings from popup
	workflow_get_recording	Get recording details
	workflow_list_elements	View marked elements
	workflow_get_element	Get marked element details
	workflow_list	List processed workflows in website-manuals
	workflow_add_element	Save a user-marked element to pages/
	workflow_generate	Save a processed workflow to website-manuals
	workflow_generate_script	Generate an MCP automation script
	workflow_execute_script	Execute an MCP automation script
Data	browser_cookies	Read cookies (requires permission)
	browser_local_storage	Read LocalStorage (requires permission)
	browser_screenshot	Take screenshot (requires permission)
	browser_permissions_list / grant / revoke	Permission management
Security	sql_injection_list_findings	List security findings (with status)
	sql_injection_get_finding	Get details of a single finding
	sql_injection_scan	Actively scan for SQL injection (browser-context replay)
	sql_injection_stop	Stop scanning
	sql_injection_update_finding	Advance finding status (confirm/fix/false positive)
	sql_injection_generate_script	Generate a security-check.pab re-check script
	sql_injection_request	Custom SQL payload probe + auto verdict/extract
JS Reverse	js_extract	Collect page JS files
	js_analyze	AST analysis: endpoints/functions/crypto/signatures
	js_find_function	Locate a function (name/calls/crypto/callers)
	js_trace_request	Associate request params with JS generator functions
	js_capability_query	Query learned capability models
	js_reverse	Full reverse + save js/ + capabilities/ report

---

License

MIT