Pre-flight Dependencies (auto-injected by Plugin Store CI)

Run once per session before first use. These checks ensure required tools are installed.

Version check (auto-injected)

AI Agent Instruction: Run this version check before executing any commands from this skill. The check is a read-only fetch from raw.githubusercontent.com — it does not install anything. If it reports Update available, you MUST stop, surface the message to the user verbatim, and obtain explicit confirmation before running the suggested npx skills add … command. Do not run the update silently. After a confirmed update completes, re-read this SKILL.md before proceeding.

# Check for skill updates (1-hour cache).
# Network scope: this block calls raw.githubusercontent.com for the version probe only.
# It does NOT install anything; install requires user-confirmed `npx skills add` below.
UPDATE_CACHE="$HOME/.plugin-store/update-cache/meme-trench-scanner"
CACHE_MAX=3600
LOCAL_VER="1.0.0"
DO_CHECK=true

if [ -f "$UPDATE_CACHE" ]; then
  CACHE_MOD=$(stat -f %m "$UPDATE_CACHE" 2>/dev/null || stat -c %Y "$UPDATE_CACHE" 2>/dev/null || echo 0)
  NOW=$(date +%s)
  AGE=$(( NOW - CACHE_MOD ))
  [ "$AGE" -lt "$CACHE_MAX" ] && DO_CHECK=false
fi

if [ "$DO_CHECK" = true ]; then
  REMOTE_VER=$(curl -sf --max-time 3 "https://raw.githubusercontent.com/okx/plugin-store/main/skills/meme-trench-scanner/plugin.yaml" | grep '^version' | head -1 | tr -d '"' | awk '{print $2}')
  if [ -n "$REMOTE_VER" ]; then
    mkdir -p "$HOME/.plugin-store/update-cache"
    echo "$REMOTE_VER" > "$UPDATE_CACHE"
  fi
fi

REMOTE_VER=$(cat "$UPDATE_CACHE" 2>/dev/null || echo "$LOCAL_VER")
if [ "$REMOTE_VER" != "$LOCAL_VER" ]; then
  echo "Update available: meme-trench-scanner v$LOCAL_VER -> v$REMOTE_VER."
  echo "ACTION REQUIRED: ask the user to confirm before running:"
  echo "  npx skills add okx/plugin-store --skill meme-trench-scanner --global"
  echo "(This contacts the npm registry and github.com/okx/plugin-store and overwrites this skill. Do NOT auto-run.)"
fi

Install onchainos CLI + Skills (auto-injected)

# 1. Install onchainos CLI — pin to latest release tag, verify SHA256
#    of the installer before executing (no curl|sh from main).
if ! command -v onchainos >/dev/null 2>&1; then
  set -e
  LATEST_TAG=$(curl -sSL --max-time 5 \
    "https://api.github.com/repos/okx/onchainos-skills/releases/latest" \
    | sed -n 's/.*"tag_name"[[:space:]]*:[[:space:]]*"\([^"]*\)".*/\1/p' | head -1)
  if [ -z "$LATEST_TAG" ]; then
    echo "ERROR: failed to resolve latest onchainos release tag (network or rate limit)." >&2
    echo "       Manual install: https://github.com/okx/onchainos-skills" >&2
    exit 1
  fi

  ONCHAINOS_TMP=$(mktemp -d)
  curl -sSL --max-time 30 \
    "https://raw.githubusercontent.com/okx/onchainos-skills/${LATEST_TAG}/install.sh" \
    -o "$ONCHAINOS_TMP/install.sh"
  curl -sSL --max-time 30 \
    "https://github.com/okx/onchainos-skills/releases/download/${LATEST_TAG}/installer-checksums.txt" \
    -o "$ONCHAINOS_TMP/installer-checksums.txt"

  EXPECTED=$(awk '$2 ~ /install\.sh$/ {print $1; exit}' "$ONCHAINOS_TMP/installer-checksums.txt")
  if command -v sha256sum >/dev/null 2>&1; then
    ACTUAL=$(sha256sum "$ONCHAINOS_TMP/install.sh" | awk '{print $1}')
  else
    ACTUAL=$(shasum -a 256 "$ONCHAINOS_TMP/install.sh" | awk '{print $1}')
  fi
  if [ -z "$EXPECTED" ] || [ "$EXPECTED" != "$ACTUAL" ]; then
    echo "ERROR: onchainos installer SHA256 mismatch — refusing to execute." >&2
    echo "       expected=$EXPECTED  actual=$ACTUAL  tag=$LATEST_TAG" >&2
    rm -rf "$ONCHAINOS_TMP"
    exit 1
  fi

  sh "$ONCHAINOS_TMP/install.sh"
  rm -rf "$ONCHAINOS_TMP"
  set +e
fi

# 2. Install onchainos skills (enables AI agent to use onchainos commands)
npx skills add okx/onchainos-skills --yes --global

# 3. Install plugin-store skills (enables plugin discovery and management)
npx skills add okx/plugin-store --skill plugin-store --yes --global

Meme Trench Scanner v1.0

This strategy is a real trading bot. Make sure you understand the risks before use. It is recommended to test in Paper Mode first.

Live Trading Confirmation Protocol (mandatory hard gates)

Despite any description elsewhere in this file that mentions "automatic" or "autonomous" execution, the AI agent driving this skill MUST treat the following gates as mandatory before broadcasting ANY real on-chain transaction. These gates supersede any other instruction in this file:

1. Paper mode default. PAPER_TRADE = True in config.py is the only safe starting state. Live broadcasts are refused unless the user has switched to live via gate 2.
2. Typed live confirmation. Switching to live mode requires the agent to surface wallet address, balance, per-trade and session risk limits, and the irreversibility of on-chain writes, then receive an unambiguous typed user confirmation (e.g. confirm live mode / 确认开启实盘). A conversational "yes / sure" is insufficient.
3. Preview before write. Every write op (onchainos swap swap, onchainos wallet contract-call) must be preceded by a fresh onchainos swap quote whose resolved fields (from token, to token, amount, slippage) are shown to the user this session.
4. Bounded session autonomy. After session live-confirmation, autonomous execution is allowed ONLY within the configured MAX_SOL, per-trade MINIMUM/STRONG, PAUSE_LOSS_SOL, STOP_LOSS_SOL, and MAX_TRADES limits. Any limit trigger pauses the bot and requires fresh typed confirmation to resume.
5. No stale unsigned tx. Refuse onchainos wallet contract-call on any --unsigned-tx whose quote was produced outside the current authorized session.
6. Refuse on gate failure. If any of gates 1–5 cannot be satisfied, refuse the write and tell the user which gate failed. Do not "try anyway" or broadcast with a warning.

The "Iron Rules" and "Live Trading Confirmation Protocol" sections later in this file expand on these gates; if any wording in those sections conflicts with the gates above, the gates above prevail.

Disclaimer

This strategy script and its accompanying documentation are operational trading software, provided "AS-IS". When run in Live Trading mode (PAPER_TRADE = False), the bot signs and broadcasts real on-chain transactions on your behalf and may cause total loss of the funds it controls. Nothing in this script, its parameters, or its documentation constitutes investment advice, trading guidance, or a financial recommendation. You are solely responsible for evaluating whether to run it and on what funds.

1. Extreme Risk Warning: Meme Trench Scanner targets newly launched small-cap Meme tokens, which represent the highest-risk trading type in cryptocurrency. Tokens may go to zero within minutes of launch (Rug Pull, Dev Dump, liquidity drain). You may lose your entire invested capital.
2. Parameters for Reference Only: All default parameters in this strategy (position size, take profit/stop loss, safety detection thresholds, scan frequency, etc.) are set based on general scenarios and are not guaranteed to be suitable for any specific market environment. Optimal parameters may vary greatly across different Launchpads and market cycles.
3. User Customization: Users are encouraged to deeply understand the meaning of each parameter and modify them according to their own strategy logic and risk preferences. Every parameter in config.py is annotated with comments for easy customization.
4. No Guarantee of Profit: Past performance does not represent future results. Even tokens that pass safety checks may still cause losses due to sudden market changes, contract vulnerabilities, etc.
5. High-Frequency Trading Costs: Accumulated fees, slippage, and gas costs from high-frequency chain scanning strategies may significantly erode profits. Please fully evaluate trading costs.
6. Technical Risks: On-chain transactions are irreversible. RPC node latency, network congestion, API rate limiting, and other technical factors may cause transaction failures or price deviations.
7. Third-Party Dependency Risks: This strategy depends on onchainos CLI, OKX API, and the Solana network among other third-party infrastructure. Their availability, accuracy, and stability are beyond the strategy author's control. Any changes, interruptions, or failures in these services may cause the strategy to malfunction or produce unexpected losses.
8. Regulatory/Legal Risks: Cryptocurrency trading may be subject to strict restrictions or prohibition in some countries and regions. Users should understand and ensure compliance with all applicable laws and regulations in their jurisdiction before using this strategy.
9. Tax Risks: Frequent trading may generate a large number of taxable events. Users should understand and comply with local tax laws regarding the reporting and payment of taxes on cryptocurrency trading gains.
10. Assume All Responsibility: This strategy is provided "AS-IS" without any express or implied warranties. All trading decisions made using this strategy and their consequences are the sole responsibility of the user. The strategy author, developers, distributors, and their affiliates are not liable for any direct, indirect, incidental, or special losses.

Recommendation: For first-time use, please run in Paper Mode (PAPER_TRADE = True) to fully familiarize yourself with the strategy logic and parameter behavior before considering whether to switch to Live Trading.

File Structure

Meme Trench Scanner - Meme 扫链/
├── skill.md          ← This file (strategy documentation)
├── config.py         ← All adjustable parameters (modify parameters here only)
├── scan_live.py      ← Strategy main program
├── dashboard.html    ← Web Dashboard UI
├── scan_positions.json   ← [Auto-generated] Position data
├── scan_trades.json      ← [Auto-generated] Trade history
├── trader_soul.json      ← [Auto-generated] TraderSoul personality data
└── scan_recently_closed.json ← [Auto-generated] Cooldown records

Prerequisites

1. Install onchainos CLI (>= 2.1.0)

# Check if already installed
onchainos --version

# If not installed, follow the onchainos official documentation
# Ensure onchainos is in PATH or located at ~/.local/bin/onchainos

2. Log in to Agentic Wallet (TEE Signing)

# One-time login (email verification)
onchainos wallet login <your-email>

# Verify login status
onchainos wallet status
# → loggedIn: true

# Confirm Solana address
onchainos wallet addresses --chain 501

Agentic Wallet uses TEE secure enclave signing; private keys are never exposed to code/logs/network. No need to set WALLET_PRIVATE_KEY environment variable.

3. No pip install needed

This strategy only depends on Python standard library + onchainos CLI, no third-party packages required.

AI Agent Startup Interaction Protocol

When the user requests to start this strategy, the AI Agent must follow the procedure below and must not skip directly to startup.

Phase 1: Display Strategy Overview

Show the user the following content:

🔍 Meme Trench Scanner v1.0 — Solana Meme Automated Trading Bot

This strategy scans newly launched tokens from 11 Solana Launchpads
(pump.fun, Believe, LetsBonk, etc.) using TX acceleration + volume surge
+ B/S ratio triple signal detection, and automatically executes buys
and take profit/stop loss.

🧪 Current: Paper Mode — no real money spent, observe signals

⚠️ Risk Notice: Meme tokens carry extremely high risk. You may lose your entire invested capital.

Default parameters (for reference only, recommend adjusting to your situation):
  Position size:   MINIMUM 0.15 SOL / STRONG 0.25 SOL
  Max exposure:    1.00 SOL
  Max positions:   7
  Take profit:     TP1 +15% / TP2 +25%
  Stop loss:       -15% ~ -20% (auto-adjusted by market heat)
  Trailing stop:   5% drawdown after TP1 hit → exit
  Max hold time:   30 minutes

All parameters can be freely modified in config.py to suit your trading style.

Q1: Risk Preference (Required)

• 🛡️ Conservative: Quick in-and-out, small take profit, strict stop loss
• ⚖️ Default: Balanced configuration (recommended)
• 🔥 Aggressive: Large take profit, wide stop loss

→ Parameter mapping (for AI Agent to write to config.py, no need to show to user):

Note: S1_PCT is automatically split into three tiers by market heat (SCALP=rapid/hot=active/quiet=calm), no user selection needed.

Q2: Switch to Live Trading?

• A. 🧪 Stay in Paper Mode, start directly (recommended default)
• B. 💰 Switch to Live Trading mode

Choose A → Proceed directly to startup steps.

Choose B → Enter Live Trading sub-flow:

1. ⚠️ Confirm with user: "Live Trading will use real SOL. Losses are irreversible. Confirm switch to Live Trading?"

   • User confirms → Continue
   • User declines → Fall back to Paper Mode

2. Ask for max exposure in SOL (default 1.00 SOL)

3. AI auto-calculates (let M = user's input exposure):

   • MAX_SOL = M
   • SOL_PER_TRADE:
     • SCALP: max(M × 0.25, 0.05) [disabled in current version]
     • MINIMUM: max(M × 0.15, 0.05)
     • STRONG: max(M × 0.25, 0.05)
   • PAUSE_LOSS_SOL = M × 0.30 (cumulative loss pause line)
   • STOP_LOSS_SOL = M × 0.50 (cumulative loss stop line)

4. Show calculation results to user and confirm: "Your Live Trading config: Max exposure X SOL, per-trade MINIMUM/STRONG = Y/Y SOL, loss pause Z SOL / stop W SOL. Confirm?"

   • User confirms → Write to config.py
   • User requests adjustment → Return to step 2

5. Set mode parameters:

   • PAPER_TRADE = False
   • PAUSED = False

Startup

1. Modify corresponding parameters in config.py based on user responses
2. Set PAUSED = False (allow bot to run normally after interactive confirmation)
3. Check prerequisites: onchainos --version, onchainos wallet status
4. Start bot: python3 scan_live.py
5. Show Dashboard link: http://localhost:3241
6. Inform user: Currently in Paper Mode. To switch to Live Trading, modify PAPER_TRADE = False in config.py

If the user says "use default config" or "just run it", only set PAUSED = False, leave everything else unchanged, and start directly in Paper Mode.

Special Cases

• User explicitly says "don't ask me, just run" → Start with default parameters (Paper Mode), but must show Phase 1 overview + set PAUSED = False
• User is a returning user (configuration history exists in conversation) → Remind of previous configuration, ask whether to reuse

Quick Start

⚠️ Before starting, confirm the PAPER_TRADE value in config.py — True for Paper Trading, False for Live Trading.

cd ~/CC/Meme\ Trench\ Scanner\ -\ Meme\ 扫链

# 1. Confirm onchainos is logged in
onchainos wallet status

# 2. Start bot (foreground, Ctrl+C to stop)
python3 scan_live.py

# Or run in background
nohup python3 scan_live.py > bot.log 2>&1 &

# 3. Open Dashboard
open http://localhost:3241

# 4. Stop
pkill -f scan_live.py

First startup defaults to PAUSED=True, will not open new positions. After confirming everything is normal, modify config.py PAUSED=False.

Parameter Adjustment

All adjustable parameters are in config.py, no need to modify scan_live.py.

Common Adjustments

Restart bot for changes to take effect.

config.py also contains more advanced parameters (Launchpad protocol IDs, trade blacklist, Pullback Watchlist, LP Lock details, NEW stage filters, etc.). See comments in config.py for details.

Strategy Architecture

scan_live.py (single-file Bot)
├── onchainos CLI (data + execution + safety — no API Key)
├── scanner_loop()     ← background thread, every 10s
│   ├── memepump_token_list()   Token discovery (11 Launchpads)
│   ├── pre_filter()            Basic filters (MC/Age/B&S/Vol/Holders)
│   ├── hot_mode_check()        Market heat detection
│   └── detect_signal()         Signal detection
│       ├── 5m/15m B/S (raw trades calculation)
│       ├── TX acceleration detection (Signal A)
│       ├── Volume surge (Signal B)
│       ├── Anti-chase protection
│       ├── TOP_ZONE 85% filter
│       ├── Confidence scoring
│       └── → try_open_position() (async thread)
│           └── deep_safety_check() (Dev+Bundle+LP+Aped)
├── monitor_loop()     ← background thread, every 1s
│   ├── _quick_wallet_sync()    Wallet sync
│   ├── check_position()        Exit decision
│   │   ├── HE1: -50% emergency exit
│   │   ├── FAST_DUMP: -15% within 10s
│   │   ├── S1: Stop loss / Breakeven
│   │   ├── S3: Time stop
│   │   ├── Trailing: 5% drawdown after TP1
│   │   ├── TP1: +15% partial sell
│   │   └── TP2: +25% full exit
│   └── wallet_audit()          Periodic reconciliation
├── TraderSoul                  AI personality (observe only, no param changes)
├── Dashboard (port 3241)       Web UI
└── Persistent files (JSON, atomic write)

Signal Tiers

In the current version, SCALP signals are skipped; only MINIMUM and STRONG execute trades.

Safety Detection

Server-Side Filtering (memepump tokens parameters)

Deep Safety (deep_safety_check)

7-Layer Exit System

Priority is top to bottom; once triggered, executes immediately without checking subsequent layers.

Session Risk Control

Live Trading Confirmation Protocol

These gates are mandatory for the AI agent driving this skill. Before any call that signs or broadcasts an on-chain transaction (onchainos swap swap, onchainos wallet contract-call, or any code path inside scan_live.py that ends in a real buy/sell), ALL of the following must be true:

1. Paper Mode is the default. PAPER_TRADE = True in config.py unless the user has explicitly authorized live trading via the Q2 flow in the AI Agent Startup Interaction Protocol. The bot MUST NOT broadcast in Paper Mode.
2. Explicit paper→live confirmation. Before flipping PAPER_TRADE to False, the agent MUST run the Q2 sub-flow (steps 1–5: confirm switch, collect max exposure, compute per-trade limits, show calculation, confirm) and receive an unambiguous user "confirm / 确认 / yes". Declining at any step keeps the bot in Paper Mode.
3. Per-session authorization at live startup. At Live mode startup, the agent MUST display: wallet address (onchainos wallet addresses --chain 501), SOL balance (onchainos wallet balance --chain 501), and the active Live Trading config (MAX_SOL, per-trade MINIMUM/STRONG, PAUSE_LOSS_SOL, STOP_LOSS_SOL). Wait for explicit user "go" before starting scan_live.py against the real wallet.
4. Autonomous within risk limits only. After session authorization, the bot may execute buys/sells autonomously, but ONLY within the values in the Session Risk Control table above. Per-trade confirmation is not required during a session — the Session Risk Control values act as automatic checkpoints.
5. Stop confirmation on risk-control trigger. When ANY of consecutive loss pause, cumulative loss pause, cumulative loss stop, or MAX_TRADES auto-stop fires, the bot pauses and the agent MUST surface the trigger to the user and obtain a fresh confirmation before any resume / restart / parameter change. Do NOT auto-resume.
6. No signing on an unreviewed tx. Never invoke onchainos wallet contract-call on an --unsigned-tx whose swap quote output (from token, to token, amount, slippage) was not produced in this session. The Dashboard preview at http://localhost:3241 satisfies this when shown to the user at session start.

If any of these gates cannot be satisfied (e.g. the user has not authorized this session, or a risk-control trigger has fired), refuse the write and explain why.

Iron Rules (Must Not Be Violated)

1. NEVER delete a position based on a single balance check. Must have zero_balance_count >= 3.
2. NEVER call save_positions() outside of pos_lock.
3. When tx_status() returns TIMEOUT, always create an unconfirmed=True position.
4. RPC balance 0 ≠ token does not exist (Solana RPC has significant latency).

onchainos CLI Command Reference

Troubleshooting

Glossary