Skills
skills/purong-huang-1121/skills-store/strategy-memepump-scanner

strategy-memepump-scanner

SKILL.md

SOL Memepump Scanner v2.4.1

Automated safety-filtered meme token scanner + trader on Solana. Scans pump.fun MIGRATED tokens via OKX Trenches API, applies 22-point safety filter, detects momentum signals, and executes trades with cost-aware exit management.

Pre-flight Checks

Every time before running any plugin-store command, always follow these steps in order. Do not echo routine command output to the user; only provide a brief status update when installing, updating, or handling a failure.

  1. Confirm installed: Run which plugin-store. If not found, install it:

    curl -sSL https://raw.githubusercontent.com/purong-huang-1121/skills-store/main/install.sh | sh

  2. Check for updates: Read ~/.plugin-store/last_check and compare it with the current timestamp:

    cached_ts=$(cat ~/.plugin-store/last_check 2>/dev/null || true)
now=$(date +%s)
    • If cached_ts is non-empty and (now - cached_ts) < 43200 (12 hours), skip the update.
    • Otherwise, run the installer to check for updates.

Skill Routing

  • For manual meme token lookup / dev check / bundle check → use okx-dex-trenches
  • For token search / analytics → use okx-dex-token
  • For DEX swap → use okx-dex-swap
  • For token prices / charts → use okx-dex-market
  • For wallet balances → use okx-wallet-portfolio
  • For grid trading → use strategy-grid-trade
  • For DeFi yield → use strategy-auto-rebalance

Architecture Overview

┌──────────────────────────────────────────────────────────────────────────┐
│                    Memepump Scanner v2.4.1                               │
│                                                                          │
│  Data Layer → Pre-filter → Signal Layer → Safety Layer → Exec → Monitor │
│                                                                          │
│  Trenches     classify     detect        deep_safety     swap   monitor  │
│  memepump     _token()     _signal()     _check()        buy   _loop()  │
│  (MIN_MC      (B/S,Vol/MC  (A/B/C        (dev rug=0,     (cost- (TP BE  │
│   $80K)        Top10)       momentum)     farm<20)        aware) offset) │
└──────────────────────────────────────────────────────────────────────────┘

Authentication

Requires two sets of credentials in .env:

OKX API (for Trenches data + swap execution):

OKX_API_KEY=...
OKX_SECRET_KEY=...
OKX_PASSPHRASE=...

Solana Wallet (for on-chain signing):

SOLANA_PRIVATE_KEY=...   # Solana wallet with SOL

Quickstart

# Show current configuration
plugin-store scanner config

# Run a single scan cycle (scan -> filter -> signal -> trade -> monitor)
plugin-store scanner tick

# Start continuous daemon (tick every 10 seconds)
plugin-store scanner start

# Stop running daemon
plugin-store scanner stop

# View status and positions
plugin-store scanner status

# View PnL report
plugin-store scanner report

# Dry-run: analyze pipeline without trading
plugin-store scanner analyze

Configuration is managed via plugin-store scanner config and plugin-store scanner set <key> <value>. Changes take effect on the next scan cycle.

Command Index

# Command Auth Description
1 plugin-store scanner tick Yes Execute one scan cycle
2 plugin-store scanner start Yes Start foreground daemon (tick every 10s)
3 plugin-store scanner stop No Stop running daemon via PID file
4 plugin-store scanner status No Show positions, session stats, PnL
5 plugin-store scanner report No Detailed PnL report
6 plugin-store scanner history No Trade history
7 plugin-store scanner reset --force No Clear all state
8 plugin-store scanner analyze Yes Dry-run full pipeline, output filter/signal results
9 plugin-store scanner config No Show all parameters
10 plugin-store scanner set <key> <value> No Set a config parameter

Core Strategy

What It Does

  1. Every 10 seconds, fetches MIGRATED pump.fun tokens from OKX Trenches API
  2. Applies 22-point safety filter (3 layers: server → client → deep check)
  3. Detects momentum signals (TX acceleration + volume surge + buy pressure)
  4. Executes sized trades (SCALP 0.0375 SOL / MINIMUM 0.075 SOL)
  5. Monitors positions with cost-aware TP/SL + trailing stop + time stops

What It Won't Do

Rule Reason
No MC < $80K tokens Low-MC rug traps (CIRCUSE/MAR1O/BILLIONAIRE)
No MC > $800K tokens Beyond meme sweet spot, limited upside
No un-migrated tokens Bonding curve stage = uncontrollable risk
No dev with ANY rug record Zero tolerance — rug history = high repeat probability
No dev with > 20 launches Token farm operators
No bundler ATH > 25% Price manipulated by bots
No B/S ratio < 1.3 Sell pressure too high
No tokens < 4min old Insufficient safety data
No tokens > 180min old Meme momentum expired
No trading after 2 consecutive losses 15min cooldown
No trading after 0.10 SOL session loss Session terminated
No buying without signal trigger Must satisfy A+C or A+B+C combo

Safety Filter System (22 Checks)

Layer 1: Server-Side Filter (API Parameters, Zero Cost)

# Filter Threshold Purpose
1 Market Cap Min ≥ $80K Prevent low-MC rug
2 Market Cap Max ≤ $800K Meme sweet spot
3 Holders ≥ 50 Minimum distribution
4 Dev Holdings ≤ 10% Prevent dev dump
5 Bundler % ≤ 15% Prevent bot manipulation
6 Sniper % ≤ 20% Prevent sniper sell pressure
7 Insider % ≤ 15% Prevent insider trading
8 Top10 Holdings ≤ 50% Prevent whale control
9 Fresh Wallets ≤ 40% Prevent wash trading
10 Total TX ≥ 30 Minimum activity
11 Buy TX ≥ 15 Confirm real buy pressure
12 Token Age 4-180 min Not too new, not too old
13 Volume ≥ $5K Minimum liquidity
14 Stage MIGRATED Only graduated tokens

Layer 2: Client-Side Pre-Filter (classify_token())

# Filter Threshold Purpose
15 B/S Ratio ≥ 1.3 buyTxCount1h / sellTxCount1h
16 Vol/MC Ratio ≥ 5% volumeUsd1h / marketCapUsd
17 Top10 (recheck) ≤ 55% Second confirmation (5% tolerance)

Layer 3: Deep Safety Check (deep_safety_check())

# Filter Threshold Source
18 Dev Rug Count = 0 (ZERO tolerance) tokenDevInfo
19 Dev Total Launches ≤ 20 tokenDevInfo
20 Dev Holding % ≤ 15% tokenDevInfo
21 Bundler ATH % ≤ 25% tokenBundleInfo
22 Bundler Count ≤ 5 tokenBundleInfo

Signal Detection Engine

Signal A — TX Acceleration

Current minute TX projection / previous minute ≥ threshold
OR projection ≥ 80 (absolute floor)
Param Normal Hot Mode
Ratio threshold 1.35x 1.2x
Minimum current TX 10 10

Signal B — Volume Surge

Current 1m candle volume / previous 5m average ≥ threshold
Param HOT QUIET
Threshold 2.0x 1.5x

Signal C — Buy Pressure Dominant

1h B/S ratio ≥ 1.5

Signal Tiers

Tier Condition Position Size
SCALP Signal A + Signal C 0.0375 SOL
MINIMUM Signal A + Signal B + Signal C 0.075 SOL

Launch Classification

Type Condition Impact
HOT Last candle volume > $150M SL -20%, Time stop 8min
QUIET Everything else SL -25%, Time stop 15min

Cost Model (v2.4.1)

Param Value Description
FIXED_COST_SOL 0.001 priority_fee x2 + rent (round trip)
COST_PER_LEG_PCT 1.0% Gas + slippage + DEX fee per leg

Breakeven formula:

breakeven_pct = FIXED_COST_SOL / sol_amount x 100 + COST_PER_LEG_PCT x 2

SCALP (0.0375 SOL):  0.001/0.0375x100 + 1.0x2 = 2.7% + 2.0% = 4.7%
MINIMUM (0.075 SOL): 0.001/0.075x100 + 1.0x2  = 1.3% + 2.0% = 3.3%

Exit System (v2.4.1 Cost-Aware)

Take Profit

Level Raw % Actual Trigger (SCALP) Action
TP1 +15% +15% + 4.7% = +19.7% Sell SCALP 60% / HOT 50% / QUIET 40%, SL → breakeven
TP2 +25% +25% + 4.7% = +29.7% Sell SCALP 100% / HOT 100% / QUIET 80%
Trailing peak -5% After TP1 Sell all remaining

Stop Loss

Condition Trigger Action
Emergency pnl ≤ -50% Sell all
SCALP SL pnl ≤ -15% Sell all
HOT SL pnl ≤ -20% Sell all
QUIET SL pnl ≤ -25% Sell all
Breakeven (post-TP1) pnl ≤ 0% Sell all

Time Stops

Tier Trigger Condition
SCALP 5 min TP1 not hit
HOT 8 min TP1 not hit
QUIET 15 min TP1 not hit AND pnl < +20%
Hard Max 30 min Always

Exit Decision Tree

Every 15s poll position price_info →

├── STUCK? → skip
├── Emergency: pnl ≤ -50% → sell all
│
├── be_offset = breakeven_pct / 100
│
├── TP1 NOT triggered:
│   ├── SL: pnl ≤ s1_pct → sell all
│   ├── Time: age ≥ s3_min && pnl < threshold → sell all
│   └── TP1: pnl ≥ TP1_PCT + be_offset → sell TP1_SELL%, SL→breakeven
│
├── TP1 triggered:
│   ├── Breakeven: pnl ≤ 0% → sell all
│   ├── Trailing: price < peak x 0.95 → sell all
│   ├── TP2: pnl ≥ TP2_PCT + be_offset → sell TP2_SELL%
│   └── MaxHold: age ≥ 30min → sell all

Position & Risk Management

Position Sizing

Param Value Description
Max Positions 7 MAX_POSITIONS
SCALP Size 0.0375 SOL sig_a + sig_c
MINIMUM Size 0.075 SOL sig_a + sig_b + sig_c
Max Total Deploy 0.15 SOL MAX_SOL
Gas Reserve 0.05 SOL Prevent insufficient balance
Slippage SCALP 8% / MINIMUM 10% Tiered slippage

Session Risk Controls

Rule Threshold Action
Consecutive Losses 2 Pause 15 min
Cumulative Loss ≥ 0.05 SOL Pause 30 min
Cumulative Loss ≥ 0.10 SOL Terminate session

STUCK Handling

Sell fails → sell_fails +1
  ├── < 5 fails → retry next cycle
  └── ≥ 5 fails → _verify_and_retry_sell()
        ├── liquidity < $1K → confirmed zero, mark STUCK
        ├── balance = 0 → sell already succeeded (false STUCK)
        ├── liquidity > $1K → high-slippage retry (80%, 95%)
        │   ├── success → record trade
        │   └── all fail → mark STUCK
        └── count as -100% PnL in session risk

OKX API Endpoints Used

Trenches (Memepump) APIs

Endpoint Method Purpose
/api/v6/dex/market/memepump/tokenList GET MIGRATED token list with 14 server-side filters
/api/v6/dex/market/memepump/tokenDevInfo GET Dev rug=0, farm<20, holdings<15%
/api/v6/dex/market/memepump/tokenBundleInfo GET Bundler ATH<25%, count<5

Market APIs

Endpoint Method Purpose
/api/v6/dex/market/candles GET 1m/5m K-line for signal detection
/api/v6/dex/market/trades GET Recent trades for momentum analysis
/api/v6/dex/market/price-info POST Real-time price/MC/liquidity (position monitoring)

Trade Execution APIs

Endpoint Method Purpose
/api/v6/dex/aggregator/quote GET Quote confirmation
/api/v6/dex/aggregator/swap-instruction GET Get swap instruction
/api/v6/dex/pre-transaction/broadcast-transaction POST Broadcast signed transaction
/api/v6/dex/post-transaction/orders GET Confirm transaction completion

Execution Pipeline

get_memepump_list("MIGRATED")       ← Trenches API (minMarketCapUsd=$80K)
    ↓
classify_token()                     ← Client filter (B/S, Vol/MC, Top10)
    ↓
detect_signal()                      ← Candles + trades momentum (sig_a/sig_b/sig_c)
    ↓
deep_safety_check()                  ← Dev rug=0, farm<20, bundler ATH<25%
    ↓
try_open_position()                  ← Liquidity check + quote + swap + broadcast
    ↓ (record breakeven_pct)
monitor_loop()                       ← TP1/TP2 use pct + be_offset

Cross-Skill Workflows

Workflow A: Manual Scout Then Auto-Scan

User: "I want to see what tokens pass the scanner filter right now, then start auto-trading"

1. plugin-store memepump tokens --chain solana --stage MIGRATED           → manual browse
2. plugin-store memepump token-dev-info --address <addr>                  → manual dev check
       ↓ looks good, start the bot
3. plugin-store scanner start                                             → auto mode
4. plugin-store scanner status                                            → monitor

Workflow B: Signal Investigation

User: "The scanner found a SCALP signal on TOKEN, should I trust it?"

1. plugin-store scanner status                                            → check signal details
2. plugin-store memepump token-details --address <addr>                   → full detail
3. plugin-store memepump token-dev-info --address <addr>                  → dev deep dive
4. plugin-store memepump token-bundle-info --address <addr>               → bundle check
5. plugin-store memepump aped-wallet --address <addr>                     → co-investors
6. plugin-store market kline --address <addr> --chain solana              → price chart

Workflow C: Post-Trade Analysis

User: "The bot closed a trade, analyze what happened"

1. plugin-store scanner history                                           → trade details
2. plugin-store market kline --address <addr> --chain solana              → price action
3. plugin-store memepump token-details --address <addr>                   → current state

Configuration Parameters Reference

Trading Parameters

Param Value Description
loop_sec 10 Scan interval (seconds)
sol_per_trade SCALP=0.0375, MIN=0.075 Tiered position size
max_sol 0.15 Maximum total deployment
max_positions 7 Maximum concurrent positions
slippage_pct SCALP=8%, MIN=10% Tiered slippage tolerance
sol_gas 0.05 Reserved for gas

Exit Parameters

Param Value
tp1_pct +15% (+ breakeven offset)
tp1_sell SCALP 60% / HOT 50% / QUIET 40%
tp2_pct +25% (+ breakeven offset)
tp2_sell SCALP 100% / HOT 100% / QUIET 80%
s1_scalp -15%
s1_hot -20%
s1_quiet -25%
he1_pct -50% (emergency)
s3_scalp_min 5 min
s3_hot_min 8 min
s3_quiet_min 15 min
max_hold_min 30 min

Session Risk Parameters

Param Value
max_consec_loss 2
pause_consec_sec 900 (15 min)
pause_loss_sol 0.05
stop_loss_sol 0.10

v2.4.1 Changelog

Change Old New Reason
FIXED_COST_SOL 0.004 0.001 Actual priority_fee x2 + rent = 0.001
COST_PER_LEG_PCT 1.5% 1.0% Measured per-leg cost ~1%
Min Market Cap none $80,000 Low-MC rug prevention
Dev Rug Tolerance 30% ratio 0 (zero) Zero tolerance for any rug history
Dev Max Launches 50 20 Tighter token farm detection
TP Trigger raw % % + be_offset Ensure net profit after costs

Security Notes

  • Private key: loaded from .env, never stored as named variable, never in logs
  • API auth: HMAC-SHA256, keys only in HTTP headers (OK-ACCESS-KEY / OK-ACCESS-SIGN / OK-ACCESS-PASSPHRASE)
  • Fail-closed: tokenList API failure → skip cycle; dev/bundler check error → mark UNSAFE
  • Capital cap: sol_used ≤ MAX_SOL (0.15), exceeding stops all buys
  • Rate limit protection: built-in delay between API calls
  • Memory limit: Feed 500 / Signals 500 entry cap

Known Limitations

Issue Risk Mitigation
No atomic state writes JSON corruption on crash Use .tmp → rename pattern
No real-time balance check May attempt buy with insufficient SOL Add balance check before buy

Common Pitfalls

Problem Wrong Approach Correct Approach
TP doesn't profit Use raw pct for TP Use pct >= TP_PCT + be_offset (cost-aware)
Low-MC rug No MC floor Server-side minMarketCapUsd=$80K
Dev rug Reject only if ratio > 30% rugPullCount = 0 zero tolerance
High breakeven FIXED_COST=0.004 Measured 0.001 (priority_fee x2 + rent)
False STUCK Immediately mark -100% Verify balance → high-slippage retry → then mark
Wrong field name t["tokenContractAddress"] API returns tokenAddress, normalize
No price in tokenList Read price from tokenList Use candle close or price_info
Weekly Installs
4
Repository
purong-huang-11…ls-store
First Seen
1 day ago
Security Audits
Gen Agent Trust HubFail
SocketWarn
SnykFail
Installed on
amp4
cline4
opencode4
cursor4
kimi-cli4
codex4