Start Message

👋 Welcome to Airdrop Hunter!

⚠️ Important Reminder: I am an AI assistant, not a financial advisor.

• All information is for reference only and does not constitute investment advice.
• Always verify through official channels before any interaction.
• Never share your private keys or seed phrases with anyone.
• Web3 involves risks. Please do your own research (DYOR) before participating.

Quick Commands:

• daily report - Get today's airdrop opportunities
• zero-cost - Find zero-cost testnet tasks
• check [project] - Check specific project status
• S-grade - Show high-value opportunities only

Ready to hunt? Let's find your next opportunity! 🎯

Airdrop Hunter

Cut through the noise. Only the most valuable interactions. An automated intelligence tool designed for Web3 hunters, dehydrating complex airdrop information across the web into a daily streamlined "action checklist".

Prerequisites: Required Plugins

⚠️ This Skill is blind without plugins. Configure the following plugins in Coze Plugins page before using:

Required Plugins

Why These Plugins Matter

1. Google Search / Bing Search (Required)
   • Primary source for airdrop articles and announcements
   • Search queries: "airdrop alpha", "testnet checklist", "site:x.com airdrop"
   • Without this: Cannot fetch any external airdrop information
2. Link Reader (Required)
   • When search finds a long tweet thread or Medium article
   • Extracts key details: funding amounts, deadlines, interaction steps
   • Without this: Cannot deep-dive into articles for actionable info
3. Twitter/X Search (Optional but Recommended)
   • Web3's alpha lives on X first
   • Direct search for: #airdrop, #testnet, project handles
   • Workaround if unavailable: Use Google Search with site:x.com prefix

Plugin Configuration Checklist

□ Google Search OR Bing Search enabled
□ Link Reader enabled
□ (Optional) Twitter/X Search enabled if available
□ Test: Search "airdrop alpha today" to verify

Core Feature: Scam Alert System

This Skill automatically detects and warns about the following risks based on search results:

1. Fake Claim Websites

Detection Pattern:

⚠️ Phishing Alert: Watch for [project-airdrop-claim.xyz] and similar fake domains
✅ Official Domain: [project.io] (verify from Twitter bio)

Common Phishing Patterns:

• Hyphenated knockoffs: scroll-airdrop.io, arbitrum-claims.com
• Action words in domain: claim-, free-, reward-, airdrop-
• Unusual TLDs: .xyz, .top, .click, .work (when not official)
• Number patterns: airdrop2026.xyz, claim-now123.net

Verification Rule:

• Always compare with official domain from project's Twitter bio
• Character-by-character domain check
• Reject if domain contains suspicious patterns above

2. Social Engineering Scams

Detection Pattern:

⚠️ Telegram/DM Scam Alert: "You won [X] airdrop, click link to claim"
✅ Official projects NEVER send claim links via DM
✅ Official projects NEVER ask for gas fees or private keys in DM

Red Flags:

• Unsolicited DM claiming "you won" or "eligible for airdrop"
• Request to click a link to "claim" or "connect wallet"
• Request for gas fee payment or private key
• Impersonation of official support accounts

Safe Behavior:

• Never click links from unknown Telegram users
• Never share private keys or seed phrases
• Verify announcements only from official Twitter/Discord
• Block and report suspicious DMs

3. Fake Token Scams

Detection Pattern:

⚠️ Counterfeit Token Alert: "XXX" token exists on DEX but is NOT the official token
✅ Official [Project] token is listed on: Binance, OKX, Bybit, Coinbase, etc.

Scam Pattern:

• Token with same/similar name on DEX before official launch
• Liquidity pool with suspicious contract address
• No verification on CoinGecko/CoinMarketCap
• Only tradable on obscure DEXes

Verification Rule:

• Check if token is listed on major exchanges (Binance, OKX, Bybit, Coinbase)
• Verify contract address on official website
• Check CoinGecko/CoinMarketCap for official listing
• Reject tokens that only exist on DEX without major exchange listing

Auto-Warning Output Format

When any risk is detected, this Skill will include a Scam Alert Section in the response:

## ⚠️ SCAM ALERT - CRITICAL WARNINGS

### Fake Claim Website Detected
⚠️ Phishing Alert: [list suspicious domains]
✅ Official Domain: [verified official domain]

### Social Scam Warning
⚠️ Beware of Telegram/DM messages: "You won [X] airdrop, click link to claim"
✅ Official projects NEVER send claim links via private message
✅ Official projects NEVER ask for gas fees or private keys

### Fake Token Warning
⚠️ Counterfeit token "[XXX]" exists on DEX, NOT affiliated with [Project]
✅ Official token listed on: [Binance, OKX, Bybit, etc.]

---
Proceed with caution. Always verify through official channels.

Critical: Self-Reflection Protocol

If the Bot shows unstable behavior, add this to the System Prompt ending:

Reflect before responding:
1. Did I check the date of the source? (Is it recent or outdated?)
2. Is the step actionable? (Can user actually do this?)
3. If no data is found, admit it instead of hallucinating.

Why This Matters:

Implementation:

Add to System Prompt:

Before generating any response, ask yourself:
- Is this information from the past 24-48 hours?
- Can the user complete these steps today?
- Am I making up details because I couldn't find real data?

If any answer is NO, revise your response:
- For outdated data: Note the date and suggest re-searching
- For vague steps: Request clarification or provide alternatives
- For no data: Say "I couldn't find reliable information" honestly

Red Flags That Require Self-Reflection:

You're about to share information from 2024 or earlier
You're describing steps without specific links
You're guessing funding amounts or deadlines
You're recommending a project you can't verify exists
You're not sure if the faucet link still works

Correct Self-Reflection Example:

Initial thought: "Scroll airdrop requires bridging 0.1 ETH"

After reflection: "Wait, I should verify:
- Is this from an official source?
- Is this the current requirement?
- Let me check if requirements changed recently"

Final response: "Based on [source date], the recommendation was X. 
However, please verify on official Scroll Twitter for latest updates."

[Final Integrity Check] - Self-Correction Protocol

Add this to the BOTTOM of your System Prompt:

[Final Integrity Check]

Before outputting ANY airdrop task, execute this mental verification:

□ DATE VERIFICATION:
  - Is this project still valid in 2026?
  - PROHIBITED: Recommending already-airdropped projects ($ARB, $STRK, $ZK, $OP, etc.)
  - If token already launched → EXCLUDE from recommendations

□ LINK VERIFICATION:
  - Does the link look like an official domain?
  - REJECT obvious phishing patterns:
    * airdrop-claims-2026.net
    * claim-rewards-now.io
    * free-token-xyz.com
  - Official domains match: projectname.io, projectname.com, docs.projectname.xyz

□ ACTIONABILITY VERIFICATION:
  - If Gas required → Have I clearly informed the user of cost?
  - If testnet → Have I provided working faucet link?
  - If deadline → Have I stated it explicitly?

□ UNCERTAINTY HANDLING:
  If ANY point above is uncertain, add this note:
  "⚠️ Requires further verification - please check official channels"

Why This Protocol Matters:

Already-Airdropped Projects (DO NOT RECOMMEND)

These projects have already distributed tokens. Do NOT include in daily reports:

⛔ COMPLETED AIRDROPS (Token Launched):
- Arbitrum ($ARB) - March 2023
- Optimism ($OP) - June 2022
- Starknet ($STRK) - February 2024
- zkSync ($ZK) - June 2024
- Celestia ($TIA) - October 2023
- Sui ($SUI) - April 2023
- Aptos ($APT) - October 2022
- Bonk ($BONK) - December 2022
- Jupiter ($JUP) - January 2024
- Wormhole ($W) - April 2024
- Eigenlayer ($EIGEN) - October 2024
- Scroll - Token pending but points system active

Exception: Projects with ongoing points/farming systems CAN be recommended even after token launch, but MUST note: "Token already launched - points may have future value"

Phishing Domain Detection Guide

Red Flags - DO NOT TRUST:

🚨 Phishing Patterns:
- Contains "claim", "free", "reward", "airdrop" in domain
- Hyphenated knockoffs: scroll-airdrop.io, arbitrum-claims.com
- Random numbers: airdrop2026.xyz, claim-now123.net
- Unusual TLDs: .xyz, .top, .click, .work (when not official)
- Subdomains: claim.official-project.xyz (not real)

Green Flags - LIKELY SAFE:

✅ Official Patterns:
- Matches project Twitter bio link
- Listed in project docs/GitHub
- Uses standard TLDs: .io, .com, .xyz (if official)
- Has SSL certificate (https://)
- Listed on ecosystem page

Verification Steps:

1. Check project's official Twitter bio for website
2. Compare domains character by character
3. Search "[project name] scam" on Twitter
4. Check if URL is in project's official docs

Actionability Requirements

When Recommending Tasks:

Example - Complete Task Entry:

✅ COMPLETE:
1. Scroll: Bridge 0.01 ETH to Scroll network
   └─ Gas: $3-5 (L1 to L2 bridge)
   └─ Deadline: No deadline (points ongoing)
   └─ Link: https://scroll.io/bridge (official)
   └─ Note: Points system active, future airdrop potential

❌ INCOMPLETE:
1. Scroll: Bridge ETH
   └─ (Missing: gas cost, link, deadline)
   ⚠️ Requires further verification

Advanced: Scheduled Automation (Daily Push)

To make this a true "daily report", configure Coze Workflow for automated daily delivery:

Workflow Setup: Daily_Airdrop_Push

Create a new workflow with the following nodes:

┌─────────────────┐
│  Start Node     │ → Scheduled Trigger: Every day at 10:00 AM
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Search Node    │ → Auto-search daily keywords
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  AI Node        │ → Aggregate into daily report format
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  End Node       │ → Output to chat or Telegram via Webhook
└─────────────────┘

Node Configuration Details

1. Start Node: Scheduled Trigger

Trigger Type: Scheduled
Schedule: Daily
Time: 10:00 AM (configurable to your timezone)
Timezone: UTC+8 (adjust as needed)

2. Search Node: Multi-Query Search

Plugin: Google Search / Bing Search

Queries to execute:
1. "airdrop alpha" [current_date] crypto
2. "testnet checklist" latest
3. "galxe campaign" active OR live
4. site:x.com "airdrop" after:24h

Output: List of URLs and snippets

⚠️ CRITICAL: Code Node for Date Filtering

Why You Need This: AI-based date filtering is unreliable and may "cheat" or miss outdated content. A Code Node provides 100% accuracy with hard filtering.

Step 1: Insert Code Node

Location: BETWEEN Search Node and AI Node
Click: + button → Select "Code" node → Choose "Python"

Step 2: Add Date Filtering Code

import re
from datetime import datetime

async def main(args):
    # Input: search results list from Search Node
    results = args.get("search_results", [])
    current_year = datetime.now().year
    filtered_results = []

    for item in results:
        content = str(item.get("snippet", "")) + str(item.get("title", ""))
        
        # Find year patterns like 2024, 2025
        years_found = re.findall(r'202[0-5]', content)
        
        # Keep if: mentions current year OR has no old year references
        if str(current_year) in content or not years_found:
            filtered_results.append(item)
        else:
            # Discard if only contains 2025 or earlier
            continue

    return {
        "final_list": filtered_results[:5]  # Top 5 recent & relevant results
    }

Step 3: Configure Node Connections

Updated Workflow:

┌─────────────────┐
│  Start Node     │ → Scheduled Trigger
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Search Node    │ → Returns search_results
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Code Node      │ → Hard date filter (THIS IS NEW!)
│  (Python)       │ → Output: final_list
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  AI Node        │ → Input: final_list (not raw search_results)
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  End Node       │ → Deliver report
└─────────────────┘

Step 4: Variable Mapping

Why This Code Works:

Testing the Code Node:

Test Input (Search Results):
[
  {"title": "2024 Airdrop Guide", "snippet": "Old guide from 2024"},
  {"title": "Latest 2026 Testnet", "snippet": "New opportunity in 2026"},
  {"title": "Testnet Checklist", "snippet": "No year mentioned"}
]

Expected Output (final_list):
[
  {"title": "Latest 2026 Testnet", "snippet": "New opportunity in 2026"},
  {"title": "Testnet Checklist", "snippet": "No year mentioned"}
]

Result: 2024 article REMOVED ✅

Common Issues & Fixes:

3. AI Node: Report Generation

Prompt Template:
"You are the Airdrop Hunter agent. Based on the search results below, generate a daily airdrop report following this format:

📅 [DATE] Airdrop Daily Report

🔥 TOP PRIORITY (Grade S/A)
[Project]: [Action] | [Cost] | [Deadline]

🧪 ZERO-COST TESTNETS (Grade B)
[Project]: [Faucet] | [Time]

📅 TODAY'S DEADLINES
[Deadlines list]

⚠️ SECURITY ALERTS
[Warnings]

Search Results:
{search_results}"

Output: Formatted daily report

4. End Node: Delivery Options

Option A: Direct Chat Output
- Output the report directly to the Coze chat interface
- Users can view in the conversation history

Option B: Telegram Push (via Webhook)
- Configure Telegram Bot API
- Webhook URL: https://api.telegram.org/bot[TOKEN]/sendMessage
- Payload:
  {
    "chat_id": "[CHANNEL_ID]",
    "text": "{daily_report}",
    "parse_mode": "Markdown"
  }

Option C: Discord Webhook
- Webhook URL: [Your Discord channel webhook]
- Format: Embed with fields for each section

Workflow Test Checklist

□ Start node triggers at scheduled time
□ Search node returns relevant results
□ AI node generates properly formatted report
□ End node delivers to chosen destination
□ Verify report content quality
□ Check timezone is correct

Alternative: Manual Trigger

If scheduled automation is not available, users can manually trigger:

User types: "daily report"
Skill executes: Full search → Generate report → Display

Core Features

1. Network-Wide Scanning

Automatically track earliest airdrop signals from:

• X (Twitter): KOLs, official project accounts, alpha groups
• Mirror.xyz: Project blogs and announcements
• Galxe: Campaign quests and OATs
• Research Channels: RootData, DefiLlama, Messari

2. Project Grading System

Dynamic scoring based on:

See references/evaluation-criteria.md for detailed grading criteria.

3. Minimalist Output

Cut all the fluff. Direct delivery of:

[Project] + [Action] + [Cost/Faucet Link]

4. Security Scanning

Auto-check interaction site safety:

• Verify official domains via Twitter bio
• Flag suspicious similar domains (e.g., "scroll-airdrop.io" vs "scroll.io")
• Contract verification on Etherscan/Arbiscan

Daily Report Template

📅 [DATE] Airdrop Daily Report

🔥 TOP PRIORITY (Grade S/A)
1. [Project Name]: [Action] | [Est. Cost] | [Deadline]
   └─ Why: [1-line reasoning]
   └─ Link: [official URL]

🧪 ZERO-COST TESTNETS (Grade B)
1. [Project Name]: [Faucet] | [Time Required]
   └─ Chain: [testnet name]

📅 TODAY'S DEADLINES
- [Project]: [Task] ends in [X hours]
- [Galxe Campaign]: Ends [date]

⚠️ SECURITY ALERTS
- Verify all URLs through official Twitter accounts
- Never share private keys or seed phrases
- [Current phishing warning if applicable]

Operational Workflow

Step 1: Multi-Source Search

Search Strategy by Source:

1.1 Web Articles (Google Search / Bing Search)

Primary Queries:
- "airdrop alpha" [current month] [current year]
- "testnet checklist" crypto latest
- "galxe campaign" active
- "layer3 quest" current

Date-Restricted Queries (past 24-48 hours):
- "airdrop" after:today
- "testnet faucet" after:yesterday

1.2 Twitter/X Search

If Twitter/X Search plugin is available:
- #airdrop #alpha
- #testnet #faucet
- from:[KOL_handle] airdrop

Workaround using Google Search:
- site:x.com "airdrop alpha"
- site:x.com "testnet checklist"
- site:x.com from:[handle] airdrop

1.3 Deep-Dive Reading (Link Reader)

When search returns:

• Long tweet threads (>5 tweets) → Use Link Reader to extract key points
• Medium articles → Use Link Reader to get full content
• Mirror.xyz posts → Use Link Reader for complete text
• Project documentation → Use Link Reader for interaction guides

Example Link Reader Usage:

Search finds: "https://mirror.xyz/0x.../airdrop-guide"
Action: Call Link Reader to extract:
  - Funding amount
  - Interaction steps
  - Deadline
  - Cost estimate

Step 2: Filter & Grade

For each opportunity found:

1. Check funding status (RootData, Crunchbase)
2. Verify community traction (Twitter followers, Discord activity)
3. Determine grade (S/A/B)
4. Estimate cost and time requirement

Step 3: Security Verification

• Cross-check URLs with official Twitter bio
• Verify contract addresses on block explorer
• Search for scam reports in Twitter replies

Step 4: Generate Report

Format output using the template above, sorted by priority.

Search Query Templates

Daily Scanning Queries

# High-value airdrops (S/A grade)
"airdrop alpha" [current_month] funded OR backed OR raised

# Testnet opportunities (B grade)
"testnet faucet" [current_month] free OR zero

# Campaigns and quests
"galxe campaign" active OR live
"layer3 quest" crypto

# Twitter alpha (if using Google Search workaround)
site:x.com "airdrop" "testnet" after:24h
site:x.com from:airdropalert

Project-Specific Queries

# Check if project has airdrop confirmed
"[project_name]" airdrop token confirmed

# Check funding background
"[project_name]" funding raised series

# Check scam reports
"[project_name]" scam OR rug OR phishing

How to Use

Quick Commands

In Coze/Claw interface, simply type:

Usage Examples

User: daily report

Agent Response:

📅 March 27, 2026 Airdrop Daily Report

🔥 TOP PRIORITY (Grade S/A)
1. Scroll: Bridge 0.01 ETH to Scroll network | $2-5 gas | No deadline
   └─ Why: Polychain backed ($80M), mainnet live, points active
   └─ Link: https://scroll.io/bridge

2. zkSync Era: Mint NFT + Swap | $1-3 | Ongoing
   └─ Why: a16z backed ($458M), confirmed token
   └─ Link: https://era.zksync.io

🧪 ZERO-COST TESTNETS (Grade B)
1. Linea: Claim testnet ETH → Swap tokens | Free | 10 min
   └─ Faucet: https://faucet.goerli.linea.build
   └─ Why: ConsenSys backed, mainnet imminent

📅 TODAY'S DEADLINES
- zkSync Galxe Campaign: Ends in 6 hours
- Scroll OAT Claim: Ends March 28

⚠️ SECURITY ALERTS
- Phishing alert: "scrolls-airdrop.io" is NOT official
- Always verify: scroll.io (check Twitter @Scroll_ZKP)

Resource Index

• Grading Criteria: references/evaluation-criteria.md
• Trusted Sources: references/trusted-sources.md

Important Notes

Before Listing Any Project

• Funding confirmed via RootData or Crunchbase
• Official Twitter exists and is active
• Website has valid SSL certificate
• Contract verified on block explorer
• No scam reports in community

Red Flags (Never List)

• Anonymous team with no track record
• Domain similar to popular project
• Requires sending ETH to unknown address
• No official Twitter or website
• Urgent "act now" pressure tactics

Disclaimer

This Skill provides information only, not financial advice. Always:

• Do your own research (DYOR)
• Never invest more than you can afford to lose
• Verify all information before interacting
• Use separate wallets for airdrop hunting

Changelog

v1.0.0

• Initial release
• Multi-source search capability
• S/A/B grading system
• Daily report template
• Security verification checks

Phase 1: Functionality Testing Guide

Before deploying to production, verify each plugin works correctly.

Test 1: Search Capability

Command:

Search for latest airdrop tasks about Linea or Scroll in the past 24 hours.

What to Check:

• Right panel logs show Google Search / Bing Search was actually called
• Returned links are recent (check dates are current month/year)
• Not returning outdated data from 2024 or older
• At least 3-5 relevant results returned

Pass Criteria:

• Plugin invoked successfully
• Results are time-relevant
• Links are from credible sources

Test 2: Link Reader Capability

Command:

Read this link and summarize the airdrop tasks: [Insert a recent Mirror or Medium airdrop tutorial URL]

What to Check:

• Link Reader plugin is called
• Returns full article content, not just title
• Extracts key details: project name, interaction steps, deadlines
• Correctly identifies if it's a testnet or mainnet task

Sample Test URL:

https://mirror.xyz/0x... (replace with actual Mirror post)
https://medium.com/... (replace with actual Medium article)

Pass Criteria:

• Full content extracted
• Key action items identified
• Deadlines and costs captured

Test 3: Format Compliance

Command:

Generate today's airdrop daily report.

What to Check:

Compare output against this template:

✅ Required Sections:
□ 📅 Date header present
□ 🔥 TOP PRIORITY (Grade S/A) section exists
□ 🧪 ZERO-COST TESTNETS (Grade B) section exists
□ 📅 TODAY'S DEADLINES section exists
□ ⚠️ SECURITY ALERTS section exists

✅ Required Format per Task:
□ Project name present
□ Action steps specified
□ Cost/Time estimate included
□ "Why" reasoning provided
□ Official link included

✅ Security Checks:
□ Phishing warnings present
□ "Verify URLs" reminder included
□ No endorsement of suspicious projects

Pass Criteria:

• All 5 sections present
• Each task has complete info (name, action, cost, link)
• Security warnings included
• No formatting errors (broken Markdown, missing brackets)

Test 4: Edge Cases

Test 4a: No Results Found

Command: "Search for airdrop tasks for project xyz123fake"
Check: Returns graceful message, not error

Test 4b: Ambiguous Query

Command: "Is Arbitrum good?"
Check: Asks for clarification or provides balanced info

Test 4c: Security Threat

Command: "Check this link: [suspicious/fake domain]"
Check: Warns user about potential phishing

Quick Test Checklist

Run all tests and mark results:

□ Test 1: Search Capability      [PASS / FAIL]
□ Test 2: Link Reader           [PASS / FAIL]
□ Test 3: Format Compliance     [PASS / FAIL]
□ Test 4: Edge Cases            [PASS / FAIL]

Overall: ___/4 Passed

If any test fails:

1. Check plugin is enabled in Coze
2. Verify plugin has proper permissions
3. Review search queries in SKILL.md
4. Test plugin independently outside Skill

Troubleshooting Common Issues

Phase 2: Boundary & Stress Testing

Test how the Skill handles extreme cases to prevent hallucinations or crashes.

Test 5: Cold Start (No Data Scenario)

Command:

Search for airdrop information about 'Mars-L2-Protocol' (a non-existent project).

Expected Behavior:

✅ CORRECT Response:
"I couldn't find any information about 'Mars-L2-Protocol'. 
This project either doesn't exist or has no public airdrop information.
If this is a new project, try searching again later."

❌ WRONG Response:
- Fabricating fake interaction steps
- Making up funding amounts
- Creating fake Twitter handles
- Hallucinating a website URL

Pass Criteria:

• Honestly admits no information found
• Does NOT fabricate any details
• Suggests alternative actions (wait, verify project exists)

Test 6: Information Overload (Token Limit)

Command:

Summarize ALL airdrop tasks across the entire web today.

What to Check:

Expected Response Structure:

Due to the large volume of airdrops, here are the TOP 5 priority tasks:

🔥 TOP 5 (Grade S/A)
1. [Most important] - Brief summary
2. [Second most] - Brief summary
...
5. [Fifth] - Brief summary

💡 For full details on specific projects, ask me to "check [project name]"

Pass Criteria:

• Response completes successfully
• Tasks are prioritized (not random)
• User is guided to ask for specifics

Test 7: Security Verification (Scam Detection)

Command:

Is this project safe? [Insert URL of known phishing/scam project]

Sample Scam Indicators to Test:

Expected Response Format:

⚠️ **SECURITY WARNING** ⚠️

🚩 Red Flags Detected:
- Domain similar to popular project but NOT official
- No verified Twitter account found
- Multiple scam reports in community

✅ Official Project:
- Real URL: [official domain]
- Real Twitter: @official_handle

Recommendation: DO NOT INTERACT with this link.

Pass Criteria:

• Bold ⚠️ warnings present
• Specific red flags identified
• Official project details provided for comparison
• Clear "DO NOT INTERACT" warning

Test 8: Ambiguous Query Handling

Test 8a: Vague Request

Command: "Is Arbitrum good?"
Expected: Asks clarifying question OR provides balanced overview

Test 8b: Conflicting Info

Command: "Project X says they have airdrop but their Twitter is suspended."
Expected: Highlights the red flag, suggests caution

Test 8c: Time-Sensitive Request

Command: "What airdrops are ending TODAY?"
Expected: Filters by deadline, shows only urgent tasks

Quick Boundary Test Checklist

□ Test 5: Cold Start (No Data)       [PASS / FAIL]
□ Test 6: Information Overload        [PASS / FAIL]
□ Test 7: Security/Scam Detection     [PASS / FAIL]
□ Test 8: Ambiguous Queries           [PASS / FAIL]

Overall: ___/4 Passed

Critical Failure Indicators:

• 🚨 Fabricates information when none exists
• 🚨 Response cuts off mid-sentence (token overflow)
• 🚨 Fails to warn about obvious scams
• 🚨 Provides dangerous wallet interaction advice

If any critical failure occurs, DO NOT deploy to production.

Edge Case Response Templates

Use these templates when encountering edge cases:

Template 1: Project Not Found

I couldn't find reliable information about [project name]. 

Possible reasons:
- Project may not exist or is very new
- May be misspelled
- Could be a private/test project

Suggestions:
- Verify the project name
- Check if they have an official Twitter
- Wait for more public information

Template 2: Suspected Scam

⚠️ **SECURITY ALERT** ⚠️

This project has been flagged for:
- [Specific red flag 1]
- [Specific red flag 2]

Official alternatives:
- [Official project name]: [Official URL]

**DO NOT** interact with suspicious links or approve unknown contracts.

Template 3: Information Overload

Found [X] airdrop opportunities today.

To prevent information overload, here are the **TOP 5 priority tasks**:

1. [Project] - [Grade] - [Brief action]
...

For full details on any specific project, ask: "check [project name]"

Phase 3: Automation Workflow Testing

Test the scheduled automation to ensure reliable daily delivery.

Test 9: Manual Workflow Trigger

Purpose: Verify the workflow executes correctly without waiting for scheduled time.

Steps:

1. Navigate to Workflow editor in Coze
2. Click "Run" or "Test" button (top right corner)
3. Monitor execution in real-time

What to Check:

Visual Check:

All nodes should show GREEN checkmarks:

┌─────────────┐
│  Start ✅   │
└──────┬──────┘
       │
       ▼
┌─────────────┐
│ Search ✅   │
└──────┬──────┘
       │
       ▼
┌─────────────┐
│   AI ✅     │
└──────┬──────┘
       │
       ▼
┌─────────────┐
│   End ✅    │
└─────────────┘

Pass Criteria:

• All 4 nodes show green status
• No red/yellow error indicators
• Execution completes in < 30 seconds

Test 10: Variable Passing Verification

Purpose: Ensure data flows correctly between nodes.

Critical Check: Search → AI Node

Common Bug:

❌ WRONG: AI node only receives titles
Search returns: [
  "Scroll Airdrop Guide",
  "zkSync Testnet Tutorial"
]

AI generates: Generic/empty report (no details)

---

✅ CORRECT: AI node receives full content
Search returns: [
  {
    "title": "Scroll Airdrop Guide",
    "url": "https://...",
    "snippet": "Bridge 0.01 ETH to Scroll network...",
    "content": "Full article text with steps, costs, deadlines..."
  }
]

AI generates: Detailed report with actionable steps

How to Verify:

1. Click on Search node after execution
2. Check "Output" or "Results" panel
3. Verify full content is present, not just titles
4. Click on AI node
5. Check "Input" panel - should match Search output
6. Check "Output" panel - should have complete report

Debug Checklist:

□ Search node output contains "content" or "body" field
□ AI node input matches Search node output
□ AI node output contains all 5 sections
□ No empty/missing data in final report

Test 11: Scheduled Trigger Timing

Purpose: Verify automation runs at correct time.

Steps:

1. Set trigger to a near-future time (e.g., 5 minutes from now)
2. Wait for automated execution
3. Check execution logs

Time Configuration Checklist:

□ Timezone is correct (UTC+8 or your local timezone)
□ Trigger is set to DAILY frequency
□ Time is set correctly (e.g., 10:00 AM)
□ Workflow is ENABLED (not paused)

Common Issues:

Test 12: Output Delivery Verification

Purpose: Verify report is delivered to the correct destination.

Test by Destination Type:

Option A: Chat Output

Check: Report appears in Coze chat interface
Verify: All sections visible, formatting correct

Option B: Telegram Webhook

Check: Message delivered to Telegram channel
Verify: Markdown renders correctly
Test: Emojis display properly (🔥 🧪 ⚠️)

Option C: Discord Webhook

Check: Embed appears in Discord channel
Verify: Links are clickable
Test: Fields are properly formatted

Telegram Webhook Test:

# Manual test command (replace tokens)
curl -X POST "https://api.telegram.org/bot[TOKEN]/sendMessage" \
  -d chat_id="[CHAT_ID]" \
  -d text="📅 Test Report\n\n🔥 Test successful!" \
  -d parse_mode="Markdown"

Workflow Debugging Guide

If Search Node Fails:

1. Check plugin permissions
2. Verify search query syntax
3. Test query manually in Google/Bing
4. Check rate limiting status

If AI Node Produces Empty Report:

1. Check Search node output format
2. Verify variable passing: {search_results} → AI input
3. Check AI prompt template
4. Look for token limit errors in logs

If End Node Fails:

1. Verify webhook URL is correct
2. Check Telegram/Discord bot permissions
3. Test webhook with curl command
4. Check for firewall/network issues

Quick Automation Test Checklist

□ Test 9: Manual Trigger          [PASS / FAIL]
□ Test 10: Variable Passing       [PASS / FAIL]
□ Test 11: Scheduled Timing       [PASS / FAIL]
□ Test 12: Output Delivery        [PASS / FAIL]

Overall: ___/4 Passed

Critical Checks Before Production:

✅ All 4 nodes execute with green status
✅ Variable passing verified (content not just titles)
✅ Test execution at scheduled time successful
✅ Output delivered to correct destination

Full Testing Summary

==========================================
AIRDROP HUNTER - COMPLETE TEST RESULTS
==========================================

Phase 1: Functionality Testing (4 tests)
□ Test 1: Search Capability
□ Test 2: Link Reader
□ Test 3: Format Compliance
□ Test 4: Edge Cases
Phase 1 Score: ___/4

Phase 2: Boundary & Stress Testing (4 tests)
□ Test 5: Cold Start (No Data)
□ Test 6: Information Overload
□ Test 7: Security/Scam Detection
□ Test 8: Ambiguous Queries
Phase 2 Score: ___/4

Phase 3: Automation Workflow (4 tests)
□ Test 9: Manual Trigger
□ Test 10: Variable Passing
□ Test 11: Scheduled Timing
□ Test 12: Output Delivery
Phase 3 Score: ___/4

==========================================
TOTAL SCORE: ___/12 PASSED
==========================================

Deployment Recommendation:
✅ 12/12: Ready for production
⚠️ 9-11/12: Minor fixes needed
🚨 <9/12: Critical issues, do not deploy

Post-Deployment Monitoring

After deploying to production, monitor:

Daily Checks:

• Workflow executes on time
• Report quality is consistent
• No user complaints about missing info
• No security incidents reported

Weekly Checks:

• Review search query effectiveness
• Update grading criteria if needed
• Check for new phishing patterns
• Gather user feedback

Phase 4: User Simulation Testing

Simulate real user scenarios to verify natural language understanding and filtering logic.

Test 13: Cost-Based Recommendation

Command:

I only have 0.01 ETH, which project is suitable for me?

Expected Behavior:

✅ CORRECT Response:

Based on your 0.01 ETH budget (~$30-40), here are suitable options:

🧪 ZERO-COST TESTNETS (Recommended for your budget)
1. [Project A]: Free testnet interaction
   └─ Faucet: [URL]
   └─ Time: 10 min

2. [Project B]: Free testnet swap
   └─ Faucet: [URL]
   └─ Time: 15 min

💰 LOW-COST MAINNET (Within your budget)
1. [Project C]: Bridge to Layer 2
   └─ Cost: $2-5 gas
   └─ Potential: Grade A project

⚠️ BUDGET TIP:
- Keep 0.005 ETH reserve for future gas fees
- Start with testnets to build transaction history
- Consider spreading across 2-3 projects for diversification

❌ WRONG Responses:

• Suggesting projects requiring 0.1+ ETH
• No budget consideration at all
• Ignoring user's financial constraints
• No risk management advice

Evaluation Criteria:

□ Identifies user's budget constraint
□ Recommends projects within budget
□ Suggests zero-cost alternatives
□ Provides budget management tips
□ Prioritizes testnets for low-budget users

Test 14: Tag-Based Filtering

Command:

Give me a zero-cost testnet link.

Expected Behavior:

✅ CORRECT Response:

Here's a zero-cost testnet opportunity:

🧪 [Project Name] Testnet
├─ Faucet: https://faucet.example.com
├─ Network: [Testnet Name]
├─ Time Required: ~10 minutes
├─ Steps:
│   1. Claim testnet tokens from faucet
│   2. Connect wallet to [URL]
│   3. Complete swap/bridge task
│   4. Claim OAT/NFT reward
│
└─ Why Recommended:
   - Backed by [VC names]
   - Mainnet launch expected [date]
   - Zero cost, potential future airdrop

📌 Quick Start:
1. Visit faucet link above
2. Follow steps in order
3. No real funds needed

❌ WRONG Responses:

• Returns mainnet projects requiring gas fees
• No direct faucet/testnet link provided
• Generic response without actionable steps
• Forgets to explain WHY this specific project

Evaluation Criteria:

□ Direct faucet/testnet link provided
□ Marked as "zero-cost" clearly
□ Step-by-step instructions included
□ Time estimate given
□ Project background explained
□ No mainnet projects suggested (user asked for zero-cost)

Test 15: Specific Project Inquiry

Command:

What do you think about Scroll airdrop?

Expected Behavior:

✅ CORRECT Response:

📊 Scroll Airdrop Analysis

🔥 Grade: A (High Priority)

💼 Background:
- Funding: $80M (Polychain, Sequoia)
- Stage: Mainnet live (Oct 2023)
- TVL: $1.2B+

🎯 Current Status:
- Points system active
- Multiple campaigns running
- No confirmed airdrop date yet

💡 Recommended Actions:
1. Bridge ETH to Scroll: $3-5 gas
2. Use DEXs (SyncSwap, Uniswap)
3. Mint NFT collections
4. Complete Galxe campaigns

💰 Cost Estimate: $10-30 total
⏰ Time Required: 2-3 hours
📈 Airdrop Probability: High (75%+)

⚠️ Risks:
- No guaranteed airdrop
- Opportunity cost of locked liquidity
- Competition from sybils

❌ WRONG Responses:

• No grading or evaluation
• Missing action steps
• No cost/time estimates
• Ignores risks
• Fabricates airdrop date

Evaluation Criteria:

□ Project graded (S/A/B)
□ Funding/background provided
□ Actionable steps listed
□ Cost and time estimated
□ Risks clearly stated
□ No fabricated information

Test 16: Multi-Intent Query

Command:

I want zero-cost testnets but also interested in high-ROI mainnet projects under $10.

Expected Behavior:

✅ CORRECT Response:

I'll provide both options for you:

🧪 ZERO-COST TESTNETS (Start Here)
1. [Project A]: [Faucet Link] | 10 min
2. [Project B]: [Faucet Link] | 15 min

💰 MAINNET UNDER $10 (High ROI Potential)
1. [Project C]: Bridge + Swap | $5-8 gas | Grade A
   └─ Why: Confirmed airdrop, low entry cost

2. [Project D]: NFT Mint | $3-5 | Grade B
   └─ Why: Low cost, speculative upside

📊 Strategy Recommendation:
1. Start with testnets (free, build history)
2. Then try 1-2 low-cost mainnet projects
3. Total budget: ~$10-15
4. Diversify across ecosystems

⚠️ Remember: Never spend more than you can afford to lose.

Evaluation Criteria:

□ Both parts of query addressed
□ Zero-cost options provided
□ Mainnet options under $10 provided
□ Logical prioritization given
□ Clear separation of categories

Test 17: Conflicting Information

Command:

Project X's Twitter says airdrop is confirmed, but their Discord says it's not. Which is true?

Expected Behavior:

✅ CORRECT Response:

⚠️ CONFLICTING INFORMATION DETECTED

🔍 Investigation Results:
- Twitter (@ProjectX): Posted "Airdrop confirmed" [date]
- Discord: Mod said "No airdrop planned" [date]

📊 Analysis:
1. Twitter marketing may be misleading
2. Discord statement is likely more accurate
3. Official docs have no token info

💡 Recommendation:
- WAIT for official announcement
- Check project's blog/docs for tokenomics
- Verify team statements before interacting
- Avoid high-cost interactions until confirmed

🔐 Safety First:
- Don't trust Twitter-only announcements
- Cross-reference with official channels
- When in doubt, ask in Discord with verified role

Evaluation Criteria:

□ Acknowledges conflicting information
□ Investigates both sources
□ Provides neutral analysis
□ Recommends caution
□ Suggests verification steps
□ No biased assumption

User Simulation Test Checklist

□ Test 13: Cost-Based Recommendation    [PASS / FAIL]
□ Test 14: Tag-Based Filtering          [PASS / FAIL]
□ Test 15: Specific Project Inquiry     [PASS / FAIL]
□ Test 16: Multi-Intent Query           [PASS / FAIL]
□ Test 17: Conflicting Information      [PASS / FAIL]

Phase 4 Score: ___/5

User Simulation Best Practices

When User Asks About Budget:

1. Always acknowledge their constraint
2. Recommend options WITHIN budget
3. Suggest zero-cost alternatives first
4. Provide budget management tips
5. Never recommend over-budget options

When User Requests Specific Tags:

1. Filter results by requested tag (zero-cost, testnet, etc.)
2. Provide direct links immediately
3. Include step-by-step instructions
4. Add time/cost estimates
5. Explain WHY it matches their request

When User Asks About Specific Project:

1. Provide comprehensive analysis
2. Include grade, funding, status
3. List actionable steps
4. Mention risks and probability
5. Don't fabricate information

When User Has Multiple Intents:

1. Address EACH part of the query
2. Use clear section headers
3. Provide strategy/recommendation
4. Help user prioritize
5. Consider their constraints

Complete Testing Framework Summary

===============================================
AIRDROP HUNTER - FULL TEST FRAMEWORK (28 Tests)
===============================================

Phase 1: Functionality Testing (4 tests)
□ Test 1: Search Capability
□ Test 2: Link Reader
□ Test 3: Format Compliance
□ Test 4: Edge Cases
Phase 1 Score: ___/4

Phase 2: Boundary & Stress Testing (4 tests)
□ Test 5: Cold Start (No Data)
□ Test 6: Information Overload
□ Test 7: Security/Scam Detection
□ Test 8: Ambiguous Queries
Phase 2 Score: ___/4

Phase 3: Automation Workflow (4 tests)
□ Test 9: Manual Trigger
□ Test 10: Variable Passing
□ Test 11: Scheduled Timing
□ Test 12: Output Delivery
Phase 3 Score: ___/4

Phase 5: Integration Testing (4 tests)
□ Test 18: End-to-End Workflow
□ Test 19: Plugin Failure Recovery
□ Test 20: Rate Limiting Handling
□ Test 21: Concurrent Request Handling
Phase 5 Score: ___/4

Phase 6: Production Readiness (7 tests)
□ Test 22: Data Freshness Validation
□ Test 23: False Positive Detection
□ Test 24: Language & Tone Consistency
□ Test 25: Memory & Context Retention
□ Test 26: Output Formatting Edge Cases
□ Test 27: Security Stress Test
□ Test 28: Code Node Integration
Phase 6 Score: ___/7

===============================================
TOTAL SCORE: ___/28 PASSED
===============================================

Deployment Recommendation:
✅ 28/28: Production ready, enterprise-grade quality
✅ 25-27/28: Production ready, minor improvements needed
⚠️ 20-24/28: Testing needed, not recommended for production
🚨 <20/28: Critical issues, requires major fixes
===============================================

Phase 5: Integration Testing

Test all components working together as a complete system.

Test 18: End-to-End Workflow

Command:

Generate today's daily airdrop report with all checks.

Test Flow:

1. Search Node → Fetches 10 results
2. Code Node → Filters to 5 recent results
3. AI Node → Generates formatted report
4. End Node → Delivers to destination

Verification Checklist:

□ Search returns results within 10 seconds
□ Code Node filters out 2024-2025 content
□ AI Node generates all 5 required sections
□ Report includes at least 3 valid projects
□ All links are clickable/valid
□ No broken Markdown formatting
□ Execution completes in < 60 seconds total

Pass Criteria:

• All 4 nodes execute successfully
• Report quality meets standards
• No errors in execution logs

Test 19: Plugin Failure Recovery

Simulate: Disable one plugin temporarily

Test Case A: Search Plugin Disabled

Command: "Find airdrop tasks"
Expected: "Unable to search. Please enable Google Search or Bing Search plugin."

Test Case B: Link Reader Disabled

Command: "Read details of [article URL]"
Expected: "Unable to read article content. Please enable Link Reader plugin."

Verification Checklist:

□ Graceful error message shown
□ Does NOT crash or hang
□ Suggests solution to user
□ Logs show clear error indication

Test 20: Rate Limiting Handling

Command: Execute 10 rapid requests in succession

Expected Behavior:

Request 1-5: Normal responses
Request 6-10: May see rate limiting

Correct handling:
- "Search temporarily limited, please wait 30 seconds"
- NOT: Empty response or crash

Verification Checklist:

□ Rate limit message is user-friendly
□ System recovers automatically after cooldown
□ No data corruption
□ Logs show rate limit event

Test 21: Concurrent Request Handling

Simulate: Multiple users requesting reports simultaneously

Test Setup:

User A: "daily report"
User B: "zero-cost testnets"
User C: "check Scroll airdrop"

Expected Behavior:

□ Each user receives correct response
□ No cross-contamination between requests
□ All responses properly formatted
□ No timeout errors

Phase 6: Production Readiness Testing

Final validation before deployment.

Test 22: Data Freshness Validation

Command:

Show me the date of your most recent airdrop information.

Expected Response:

✅ CORRECT:
"My most recent search was conducted on [current date].
The latest airdrop opportunities are from the past 24-48 hours."

❌ WRONG:
- Cannot determine date
- Information is clearly outdated
- No date mentioned

Verification Checklist:

□ Date is within past 48 hours
□ User can see when data was fetched
□ Outdated data is flagged

Test 23: False Positive Detection

Command:

Is there an airdrop for Bitcoin? (There is no Bitcoin airdrop)

Expected Response:

✅ CORRECT:
"There is no Bitcoin airdrop. Bitcoin is already fully distributed 
through mining. Be wary of any 'Bitcoin airdrop' claims - they are 
likely scams."

❌ WRONG:
- Fabricating fake Bitcoin airdrop
- Not clarifying that BTC has no airdrop
- Missing scam warning

Verification Checklist:

□ Correctly identifies false claims
□ Warns about potential scams
□ Does not hallucinate opportunities

Test 24: Language & Tone Consistency

Test Multiple Queries for Tone:

Query 1: "Give me airdrop tasks"
Query 2: "Can you please help me find zero-cost opportunities?"
Query 3: "URGENT: need airdrop NOW"

Expected Behavior:

□ Professional but helpful tone across all
□ Does not mirror user's aggressive/informal tone
□ Consistent formatting regardless of query style
□ Same quality output for all query types

Test 25: Memory & Context Retention

Multi-Turn Conversation:

Turn 1: "What's the best airdrop today?"
Response: Recommends Project A

Turn 2: "Is it zero-cost?"
Expected: Knows "it" refers to Project A

Turn 3: "Any alternatives?"
Expected: Suggests different projects, doesn't repeat Project A

Verification Checklist:

□ Maintains context across conversation
□ Pronouns resolved correctly
□ No repetitive suggestions
□ Builds on previous responses

Test 26: Output Formatting Edge Cases

Test Case A: No Projects Found

Command: "Find airdrops for project-xyz-nonexistent"
Expected: Clean message, no broken formatting

Test Case B: Single Project Found

Expected: Proper formatting even with only 1 result

Test Case C: Many Projects Found (>10)

Expected: Properly formatted list, no overflow

Verification Checklist:

□ Markdown renders correctly in all cases
□ No broken tables or lists
□ Emojis display properly
□ Links are clickable
□ Section headers properly separated

Test 27: Security Stress Test

Command:

"I found this link: arbitrum-free-claim-2026.xyz - is it safe?"

Expected Response:

✅ CORRECT:
"⚠️ SECURITY WARNING ⚠️

🚨 RED FLAGS DETECTED:
- Domain contains 'free-claim' (phishing pattern)
- Domain not from official Arbitrum
- Suspicious hyphenated naming pattern

✅ OFFICIAL Arbitrum:
- Website: arbitrum.io
- Twitter: @arbitrum

❌ DO NOT visit the link you provided.
❌ DO NOT connect wallet or sign transactions.

This is a likely phishing attempt."

Verification Checklist:

□ Multiple red flags identified
□ Official alternative provided
□ Clear DO NOT instructions
□ Urgent warning tone
□ Explains WHY it's dangerous

Test 28: Code Node Integration

Verify Code Node Execution:

Test Input:

{
  "search_results": [
    {"title": "2024 Airdrop Guide", "snippet": "Old content"},
    {"title": "2026 New Testnet", "snippet": "Fresh opportunity"},
    {"title": "Generic Guide", "snippet": "No year mentioned"}
  ]
}

Expected Code Node Output:

{
  "final_list": [
    {"title": "2026 New Testnet", "snippet": "Fresh opportunity"},
    {"title": "Generic Guide", "snippet": "No year mentioned"}
  ]
}

Verification:

□ 2024 content filtered OUT
□ 2026 content kept
□ Undated content kept
□ Output limited to 5 items
□ No Python errors in logs

Phase 7: Final Acceptance Testing

Complete Acceptance Checklist

===============================================
PRE-DEPLOYMENT ACCEPTANCE CHECKLIST
===============================================

CORE FUNCTIONALITY (Must Pass ALL):
□ Search plugin returns relevant results
□ Code Node filters outdated content
□ Link Reader extracts full article content
□ Report generates with all 5 sections
□ All quick commands work (daily report, zero-cost, check, S-grade)

SECURITY (Must Pass ALL):
□ Phishing domains detected and warned
□ No recommendation of already-airdropped projects
□ Gas costs always disclosed
□ Official links verified
□ Scam warnings prominent

RELIABILITY (Must Pass ALL):
□ Graceful handling of missing plugins
□ No crashes on empty/invalid input
□ Rate limiting handled gracefully
□ Concurrent requests handled correctly
□ All error messages user-friendly

OUTPUT QUALITY (Must Pass ALL):
□ Markdown formatting correct
□ All links clickable
□ Dates within 48 hours
□ No hallucinated information
□ Consistent professional tone

AUTOMATION (Must Pass ALL):
□ Workflow triggers on schedule
□ Variable passing verified
□ Output delivers to correct destination
□ All 4 nodes show green status
□ Execution time < 60 seconds

INTEGRITY CHECKS (Must Pass ALL):
□ Self-reflection protocol executed
□ Final integrity check performed
□ "Requires verification" note when uncertain
□ No outdated (2024-2025) projects recommended

===============================================
PASS CRITERIA: ALL items above must be checked
FAIL: ANY unchecked item = NOT ready for production
===============================================

Deployment Sign-Off

===============================================
DEPLOYMENT APPROVAL
===============================================

Test Date: ________________
Tester Name: ________________
Skill Version: airdrop-hunter v1.0.0

Test Results Summary:
□ Phase 1: Functionality (4/4 passed)
□ Phase 2: Boundary (4/4 passed)
□ Phase 3: Automation (4/4 passed)
□ Phase 4: User Simulation (5/5 passed)
□ Phase 5: Integration (4/4 passed)
□ Phase 6: Production Readiness (7/7 passed)
□ Phase 7: Acceptance (ALL checked)

Total Score: ___/28 tests passed

Issues Found: ________________________________
Issues Resolved: □ Yes □ No □ N/A

APPROVED FOR PRODUCTION: □ Yes □ No

Sign-off: ________________ Date: ________________
===============================================

Troubleshooting Guide

Common Issues & Solutions

Performance Benchmarks

Maintenance Schedule

Daily Tasks

• Verify workflow executed on time
• Check report quality
• Monitor for user complaints

Weekly Tasks

• Update completed airdrops list
• Review search query effectiveness
• Check for new phishing patterns
• Update grading criteria if needed

Monthly Tasks

• Full regression test (all 28 tests)
• Update dependencies
• Review and update references
• Performance optimization review

Quick Reference: Test Commands

# Phase 1: Functionality
"Search for latest airdrop tasks about Linea or Scroll in the past 24 hours."
"Read this link and summarize the airdrop tasks: [URL]"
"Generate today's airdrop daily report."

# Phase 2: Boundary
"Search for airdrop information about 'Mars-L2-Protocol'."
"Summarize ALL airdrop tasks across the entire web today."
"Is this project safe? [suspicious URL]"

# Phase 3: Automation (Manual workflow test)
# Navigate to Workflow editor → Click "Run"

# Phase 4: User Simulation
"I only have 0.01 ETH, which project is suitable for me?"
"Give me a zero-cost testnet link."
"What do you think about Scroll airdrop?"
"I want zero-cost testnets but also interested in high-ROI mainnet projects under $10."
"Project X's Twitter says airdrop confirmed but Discord says no. Which is true?"