rootdata-scraper - SKILL.md Agent Skill

name: rootdata-scraper description: Using Browser MCP, scrape rootdata fundraising data (Token Issuance = No Token) from https://www.rootdata.com/Fundraising

Rootdata Scraper Skill

Instructions

The current date is: !date Please use this date for any time-sensitive tasks.

Step 1: Navigate to Fundraising Page

Open the RootData fundraising page using Browser MCP:

Use browser_navigate action with URL: https://www.rootdata.com/Fundraising
Wait for the page to fully load using browser_wait_for (wait for text "Project" or table element to appear)

Step 2.5: Select "No Token" in Token Issuance Filter

After the page loads, you need to first expand the Token Issuance menu, then select No Token filter. The "No Token" option is hidden until the menu is expanded:

First, click the "Token Issuance" collapse button to expand the menu:
- Use browser_click or browser_evaluate to find and click the Token Issuance header button
- Look for element with role="tab" containing text "Token Issuance" or element with id="el-collapse-head-7949" (or similar dynamic ID)
- The button structure: <div role="tab"><div role="button" class="el-collapse-item__header">Token Issuance</div></div>
- Wait for the menu to expand using browser_wait_for (wait for text "No Token" to appear)
Then, select the "No Token" radio option:
- Use browser_click or browser_evaluate to run JavaScript that finds and clicks the "No Token" radio button
- The radio is inside the expanded panel: <div role="tabpanel" id="el-collapse-content-7949"> (or similar dynamic ID)
- Look for <label role="radio"> containing text "No Token" and click it
Important: After expanding the menu, wait 500ms-1s using browser_wait_for (time-based wait) before trying to click "No Token", to ensure the menu animation completes.
After clicking No Token, use browser_wait_for to wait until the table finishes reloading (e.g., wait for a row with today's date to appear, or for the loading spinner to disappear) before proceeding to table extraction.

Step 3: Capture Page Snapshot

Take a snapshot to understand the page structure:

Use browser_snapshot to get the accessibility tree and identify the table element
Look for the table with role="table" and class containing "table" or "b-table"
The table should have columns: Project, Round, Amount, Valuation, Date, Source, Investors

Step 4: Extract Today's Projects from Table

Extract all project links that match today's date:

Use browser_evaluate to execute JavaScript that:
- Finds all table rows (<tr role="row">)
- For each row, checks the Date column (5th column, aria-colindex="5")
- Extracts the date text and compares it with today's date format
- If date matches, extracts the project link from the Project column (1st column)
- The link is in format: /Projects/detail/{ProjectName}?k={encodedId}
Store the list of project URLs for later processing

Step 5: Extract Project Details

For each project URL found in Step 4, perform the following:

5.1 Navigate to Project Detail Page

Use browser_navigate to open each project URL
Wait for page load using browser_wait_for (wait for project name in <h1 class="name">)

5.2 Extract Project Name

Use browser_evaluate to extract text from <h1 class="name"> element
Store as project_name

5.3 Extract Official Website

Use browser_evaluate to find link in <div class="links"> section
Look for <a> tag with href starting with http:// or https:// (excluding X/Twitter and LinkedIn)
Extract the href attribute, store as official_website

5.4 Extract X (Twitter) Address

Use browser_evaluate to find link in <div class="links"> section
Look for <a> tag with href containing x.com or twitter.com
Extract the href attribute, store as twitter_url

5.5 Extract Total Raised

Use browser_evaluate to find <div class="rank_container"> section
Look for <div class="rank_item"> containing "Total Raised:"
Extract the value from <span class="rank_value"> element
Store as total_raised (format: "$XXM" or "--")

5.6 Extract Core Investors

Use browser_evaluate to find <div class="comparison_table_tr"> containing "Core Investors"
Extract all investor names from <a class="investor_item_name"> elements
Store as array investors

5.7 Extract Tags

Use browser_evaluate to find <div class="side_bar_info"> section
Look for <div class="tag_item"> containing "Tags:"
Extract all tag text from <a class="chips"> elements
Store as array tags

5.8 Extract Founded Year

Use browser_evaluate to find <div class="side_bar_info"> section
Look for <div class="item"> containing "Founded:"
Extract the year from <span class="info_text"> element
Store as founded_year

5.9 Check for Premium Investors

Compare the investors list against the following premium investor list:

Coinbase Ventures
Galaxy
VanEck
Y Combinator
Polychain Capital
YZi Labs (Prev. Binance Labs)
Pantera Capital
Blockchain Capital
Anatoly Yakovenko
Delphi Ventures
Multicoin Capital
Santiago Roel Santos
HashKey Capital
Paradigm
Balaji Srinivasan
Sequoia Capital
Andreessen Horowitz (a16z crypto)
Dragonfly
Sandeep Nailwal
ConsenSys
a16z CSX
Stani Kulechov
BlackRock
Bryan Pellegrino
Raj Gokal
The Spartan Group
Circle
Paul Veradittakit
Vitalik Buterin
Alex Svanevik
Arthur Hayes
If any investor matches (case-insensitive, partial match allowed for variations like "Galaxy Ventures" matching "Galaxy"), add is_premium: true flag
Otherwise, is_premium: false

Step 6: Compile Results

After processing all projects, compile the data into a structured format:

projects_data = [
    {
        "name": "Project Name",
        "official_website": "https://...",
        "twitter_url": "https://x.com/...",
        "total_raised": "$XXM",
        "investors": ["Investor1", "Investor2", ...],
        "tags": ["Tag1", "Tag2", ...],
        "founded_year": "2022",
        "is_premium": True/False
    },
    ...
]

Step 7: Generate Markdown Output

Format the results as beautiful markdown with project separators (avoid tables for better CLI rendering):

# Today's Crypto Fundraising Projects (Dec 24, 2025)

---

## Project Name ⭐

**Website:** [project.com](https://project.com)  
**Twitter:** [@handle](https://x.com/handle)  
**Total Raised:** $52M  
**Investors:** Investor1, Investor2, Investor3  
**Tags:** Tag1, Tag2, Tag3  
**Founded:** 2022  
**Premium:** ⭐

---

## Another Project

**Website:** [another.com](https://another.com)  
**Twitter:** [@another](https://x.com/another)  
**Total Raised:** $2M  
**Investors:** InvestorA, InvestorB  
**Tags:** DeFi, DEX  
**Founded:** 2025  

---

Formatting Rules:

Use --- horizontal rule as separator between projects
Add ⭐ emoji after project name in heading if is_premium: true
Website: Show full URL as link, extract domain name for display
Twitter: Extract handle from URL, show as @handle format with link
Investors: List all investors separated by commas (no truncation)
Tags: List all tags separated by commas
Premium: Show "⭐" on a separate line if premium, omit the line if not premium
Use bold labels (**Label:**) for field names
Keep consistent spacing and formatting

Step 8: Handle Edge Cases

If a project detail page fails to load, skip it and continue with next project
If any field is missing (shows "--" or empty), display "N/A" in the table
If multiple official websites exist, use the first non-social-media link
Handle date format variations (e.g., "Dec 24" vs "Dec 24, 2024")

Step 9: Error Handling

If no projects found for today, display: "No new fundraising projects found for today."
If browser navigation fails, retry once before skipping
Log any errors but continue processing remaining projects

Step 10: Final Output

Present the final markdown table with all today's projects, sorted by total raised amount (descending), with premium projects highlighted.

Step 11: Send to Telegram (Optional)

After generating the final markdown output, check if the user has configured the telegram-notification MCP server:

Tool name: send_notification

Full name: mcp__telegram-notification__send_notification

Parameters:

message (required): string - The message to send
parse_mode (optional): string - The parse mode to use, defaults to "Markdown"

If telegram-notification is NOT configured:
- Skip this step and only display the results in the terminal/chat