Onchain Alpha Radar

OnChain Alpha Radar

A systematic on-chain data research pipeline that transforms raw blockchain signals into actionable research reports.

Pipeline Overview

Token Discovery → Holdings Analysis → Smart Money Tracking → Research Report
     ↓                  ↓                    ↓                    ↓
  onchainos          onchainos            onchainos          deep-research
  + Twitter          + Twitter            + Twitter          + excalidraw

Tool Dependencies

This skill orchestrates four external tool sources:

Before starting, verify tool availability:

1. Run which onchainos — if missing, install via curl -sSL https://raw.githubusercontent.com/okx/onchainos-skills/main/install.sh | sh
2. Check if Twitter MCP tools are available (e.g., search_twitter)
3. Check if excalidraw diagram skill is present at .claude/skills/excalidraw-diagram/

If any tool is unavailable, inform the user and proceed with what IS available. The pipeline degrades gracefully — each phase can work independently.

Phase 1: Token Discovery

Goal: Identify high-potential tokens through on-chain signals and social buzz.

Step 1.1: On-Chain Signal Scan

Choose the appropriate discovery method based on user intent:

A. Trending Token Scan (broad discovery):

# Top tokens by volume in last 24h
onchainos token trending --chains <target-chains> --sort-by 5 --time-frame 4

# Top tokens by price change in last 1h (momentum)
onchainos token trending --chains <target-chains> --sort-by 2 --time-frame 2

B. Meme Token Discovery (degen alpha):

# New launches on Solana (pumpfun)
onchainos market memepump-tokens 501 --stage NEW

# Recently migrated (survived bonding curve)
onchainos market memepump-tokens 501 --stage MIGRATED

C. Specific Token Lookup (user provides name/address):

onchainos token search <query>
onchainos token price-info <address> --chain <chain>

Step 1.2: Social Signal Cross-Reference

For each interesting token found in Step 1.1, gather Twitter sentiment:

# Search for token mentions
search_twitter(keywords="$TOKEN_SYMBOL OR $TOKEN_NAME", min_likes=10, limit=20)

# Advanced search with time filter
search_twitter_advanced(keywords="$TOKEN_SYMBOL", min_likes=50, since_date="YYYY-MM-DD", product="Top")

Step 1.3: Discovery Scoring

Rate each discovered token on a 1-5 scale across:

Output a ranked shortlist of 3-10 tokens for deep analysis.

Phase 2: Holdings Analysis

Goal: Deep-dive into token economics and holder distribution.

Step 2.1: Token Fundamentals

For each shortlisted token:

# Full price and market data
onchainos token price-info <address> --chain <chain>

# Holder distribution (top 20)
onchainos token holders <address> --chain <chain>

# K-line data for price trend
onchainos market kline <address> --chain <chain> --bar 1H

Step 2.2: Meme Token Due Diligence (if applicable)

# Developer reputation
onchainos market memepump-token-dev-info <address> --chain <chain>

# Bundle/sniper detection
onchainos market memepump-token-bundle-info <address> --chain <chain>

# Audit details
onchainos market memepump-token-details <address> --chain <chain>

# Other tokens by same dev
onchainos market memepump-similar-tokens <address> --chain <chain>

Step 2.3: Key Wallet Holdings

For significant holder addresses discovered in Step 2.1:

# What else does this whale hold?
onchainos portfolio all-balances --address <whale-address> --chains 1,56,501,8453

# Total portfolio value
onchainos portfolio total-value --address <whale-address> --chains 1,56,501,8453

Step 2.4: Twitter Profile of Key Holders (if identifiable)

# If wallet is linked to a known Twitter account
get_twitter_user(username="whale_account")
get_twitter_user_tweets(username="whale_account", limit=10)
get_twitter_kol_followers(username="whale_account")

Step 2.5: Holdings Summary

Produce a structured analysis:

## Holdings Analysis: $TOKEN

### Token Economics
- Market Cap: $X | FDV: $Y
- Liquidity: $Z | Liquidity/MCap: N%
- Holders: N | Top 10 Concentration: X%

### Holder Distribution
| Rank | Address (short) | % Supply | Wallet Type |
|------|----------------|----------|-------------|

### Risk Assessment
- Dev Holdings: X%
- Bundle/Sniper %: X%
- Insider %: X%
- Rug Pull History: X/Y tokens

### Price Trend (24h)
- 5min: +X% | 1h: +X% | 4h: +X% | 24h: +X%
- Volume 24h: $X | Txs 24h: N

Phase 3: Smart Money Tracking

Goal: Identify what smart money, whales, and KOLs are buying/selling.

Step 3.1: Signal Collection

# Smart money buy signals (type 1)
onchainos market signal-list <chain> --wallet-type 1 --min-amount-usd 10000

# Whale signals (type 3)
onchainos market signal-list <chain> --wallet-type 3 --min-amount-usd 50000

# KOL/Influencer signals (type 2)
onchainos market signal-list <chain> --wallet-type 2

For a specific token:

onchainos market signal-list <chain> --token-address <address> --wallet-type 1,2,3

Step 3.2: Wallet Deep Dive

For each smart money wallet address:

# Full portfolio
onchainos portfolio all-balances --address <sm-wallet> --chains 1,56,501,8453

# Recent trades on the token
onchainos market trades <token-address> --chain <chain>

Step 3.3: Aped Wallet Network (Meme Tokens)

# Who else is holding what smart money holds?
onchainos market memepump-aped-wallet <token-address> --chain <chain>

Step 3.4: Social Intelligence on Smart Money

# Search for wallet address mentions
search_twitter(keywords="<wallet-address-short>", limit=10)

# Search for token + smart money narrative
search_twitter_advanced(keywords="$TOKEN smart money OR whale", min_likes=20, product="Top")

# Check deleted tweets (alpha leak detection)
get_twitter_deleted_tweets(username="suspected_kol", limit=20)

Step 3.5: Smart Money Flow Map

Synthesize findings into:

## Smart Money Flow: $TOKEN

### Signal Summary
| Wallet Type | # Wallets | Total USD | Avg Sold % |
|-------------|-----------|-----------|------------|
| Smart Money | N | $X | Y% |
| Whale | N | $X | Y% |
| KOL | N | $X | Y% |

### Key Wallets
| Address | Type | Amount | Still Holding? | PnL |
|---------|------|--------|----------------|-----|

### Conviction Score
- Smart money buying + low sold ratio = HIGH conviction
- Mixed signals = MEDIUM conviction
- Smart money selling = LOW conviction

Phase 4: Research Report Output

Goal: Synthesize all findings into a professional research report with visualizations.

Step 4.1: Data Consolidation

Aggregate all data from Phases 1-3 into structured sections:

1. Executive Summary — 1-paragraph verdict with conviction rating
2. Token Overview — fundamentals, economics, price action
3. On-Chain Analysis — holder distribution, whale activity, liquidity depth
4. Smart Money Activity — signal analysis, wallet tracking, conviction assessment
5. Social Sentiment — Twitter buzz, KOL mentions, community growth
6. Risk Assessment — rug pull flags, bundle detection, dev history, honeypot check
7. Conclusion & Recommendation — actionable insights with risk/reward framing

Step 4.2: Generate Excalidraw Diagrams

Create visual diagrams to accompany the report. Use the excalidraw-diagram skill to generate:

Diagram 1: Token Flow Architecture

• Show token movement between key wallets (dev, top holders, smart money, DEX liquidity)
• Use fan-out pattern for distribution, convergence for accumulation

Diagram 2: Smart Money Signal Timeline

• Timeline pattern showing when smart money entered/exited
• Overlay with price chart key levels

Diagram 3: Risk Assessment Matrix

• Side-by-side comparison of risk factors
• Color-coded: green (safe), yellow (caution), red (danger)

Diagram 4: Holder Distribution

• Visual representation of token concentration
• Highlight dev/insider/bundler portions

Step 4.3: Report Generation

Apply the deep-research methodology for rigorous analysis:

• Source Tiering: On-chain data = L1 (primary source), Twitter data = L3-L4 (supporting evidence)
• Fact Cards: Each claim links to specific on-chain transaction or data point
• Confidence Levels: Mark each conclusion with HIGH/MEDIUM/LOW confidence
• Time Sensitivity: Crypto data has extremely short shelf life — flag data staleness

Step 4.4: Report Format

Output as a Markdown file with this structure:

# Alpha Radar Report: $TOKEN_SYMBOL ($TOKEN_NAME)

> **Date**: YYYY-MM-DD | **Chain**: X | **Analyst**: OnChain Alpha Radar
> **Conviction**: HIGH/MEDIUM/LOW | **Risk Level**: 1-5

## TL;DR
[One-sentence verdict]

## 1. Token Overview
[From Phase 1 + 2 data]

## 2. On-Chain Deep Dive
[From Phase 2 data, including holder analysis]

## 3. Smart Money Activity
[From Phase 3 data]

## 4. Social Sentiment
[Twitter data aggregation]

## 5. Risk Assessment
[Comprehensive risk table]

## 6. Visual Analysis
[Reference generated Excalidraw diagrams]

## 7. Conclusion
[Actionable recommendation with explicit risk/reward]

---
*Data sources: OKX OnchainOS API, Twitter/X, On-chain transactions*
*This is not financial advice. DYOR.*

Execution Modes

The user can trigger the full pipeline or individual phases:

Full Pipeline

"Research $TOKEN for me" / "Full alpha report on $TOKEN" / "给我做一份 $TOKEN 的研报"

Run all 4 phases sequentially, generating a complete report.

Individual Phases

"What's trending on Solana?" → Phase 1 only "Analyze holders of $TOKEN" → Phase 2 only "Track smart money on $TOKEN" → Phase 3 only "Generate a report from my findings" → Phase 4 only

Quick Scan Mode

"Quick scan $TOKEN" / "快速扫描 $TOKEN"

Abbreviated pipeline:

1. Token price-info + basic fundamentals
2. Top 5 holders check
3. Smart money signal check (if any)
4. 1-paragraph summary (no diagrams)

Chain Reference

Common chain indices for onchainos commands:

Default to Solana (501) and Ethereum (1) if user doesn't specify a chain.

Error Handling

• If onchainos returns empty data: try alternate chains or broaden search parameters
• If Twitter search returns nothing: try shorter/simpler keywords, remove filters
• If a token address isn't found: use onchainos token search by symbol first, then retry with the discovered address
• If rate limited: wait 5 seconds and retry once, then inform user
• Always inform the user which data was successfully collected and which was unavailable