GMGN Agent Skills & CLI

The --condition-orders parameter allows you to attach up to 10 sub-orders (take-profit or stop-loss) to a successful swap. These orders are executed as a best-effort strategy.

Sub-order Types (order_type)

• profit_stop: Fixed take-profit.
• loss_stop: Fixed stop-loss.
• profit_stop_trace: Trailing take-profit.
• loss_stop_trace: Trailing stop-loss.

Key Fields

• side: Always set to "sell".
• price_scale:
  • For profit_stop / profit_stop_trace: The gain % from entry (e.g., "100" = +100% / 2x entry).
  • For loss_stop / loss_stop_trace: The drop % from entry (e.g., "65" = drops 65%, triggers at 35% of entry).
• sell_ratio: Percentage of position to sell when triggered (e.g., "100" for 100%).
• drawdown_rate: (Required for _trace types) The trailing callback %. E.g., "50" means the order fires if the price falls 50% from its peak.

Sell Ratio Modes (--sell-ratio-type)

• buy_amount (default): Sells a fixed percentage of the original amount bought. If you buy 100 tokens and set two 50% profit stops, each stop sells 50 tokens.
• hold_amount: Sells a percentage of the current position held at trigger time. This includes any tokens added to the position since the initial swap.

[
  {"order_type": "profit_stop", "side": "sell", "price_scale": "100", "sell_ratio": "100"},
  {"order_type": "loss_stop",   "side": "sell", "price_scale": "60",  "sell_ratio": "100"}
]

The gmgn-market skill provides several key metrics and abstractions for analyzing tokens:

Chain-specific Safety Filters

When the --filter flag is omitted, default filters are applied automatically:

• SOL: Defaults to renounced frozen (mint and freeze authority renounced).
• EVM (BSC / Base / ETH): Defaults to not_honeypot verified renounced. Note: Omitting --filter is not the same as 'no filter'. To use a custom set, you must explicitly specify all desired tags.

Key Risk and Bullish Metrics

• rug_ratio: A 0–1 score (values > 0.3 are high-risk) estimating rug pull likelihood.
• smart_degen_count / renowned_count: Number of GMGN-tagged smart money (smart_degen) or KOL (renowned) wallets holding/trading the token.
• is_honeypot: (EVM only) Indicates if the contract prevents selling. On SOL, this field is always empty/null; do not interpret null as 'safe'.
• cto_flag: 1 indicates a Community Takeover (original dev abandoned the project).
• wash_trading / bundler_rate: High values (> 0.3) suggest artificial volume or bot manipulation.
• creator_token_status: creator_hold (dev holds tokens, risk of sell pressure) vs creator_close (dev sold/burned, exit signal confirmed).

Volume vs Amount (K-line)

• volume: The USD dollar value of trades.
• amount: The number of token units traded. Example: For a token at $0.0002, a $1,214 volume corresponds to 6,070,000 amount.

The gmgn-swap skill allows for executing real, irreversible blockchain transactions on Solana, BSC, Base, or Ethereum.

Key Concepts:

• Smallest Unit: The --amount flag always requires the token's smallest indivisible unit (e.g., lamports for SOL, or 18 decimals for most EVM tokens). Do not use human-readable amounts.
• Slippage: An integer (0–100) representing price tolerance. Use --auto-slippage for volatile tokens to let GMGN set this automatically.
• Amount vs Percent: --amount (exact quantity in smallest units) and --percent (percentage of balance) are mutually exclusive. Note: --percent cannot be used if the input_token is a currency token (SOL, BNB, ETH, USDC).
• Anti-MEV: Protects against frontrunning/sandwich attacks. Enabled by default, but not supported on the base chain.
• Authentication: swap and most order commands require both GMGN_API_KEY and GMGN_PRIVATE_KEY. The private key is used for local signing and never leaves your machine.
• Order Status: After a swap, use order get --order-id to poll for status. A transaction is only successful when the status is confirmed.

The gmgn-holder-analysis skill produces a structured report to evaluate token holder behavior and risk. The output is categorized into three main sections:

1. Holder Buying Power (💰 Holder Buying Power)

Measures the remaining liquidity (ammo) available to current holders for adding to their positions. Wallets are classified by their native USD balance:

• Zero balance (⚫): No buying power; likely sub-wallets used for distribution.
• Low (<$200) (🟡): Minimal buying power.
• Mid ($200~$1200) (🟠): Moderate buying power.
• High ($1200+) (🔴): Significant buying power; can add to positions at any time.

2. Chip Structure (📊 Chip Structure)

Analyzes the distribution and health of token holdings:

• Profit/Loss Distribution: Percentage of holders in profit, loss, or break-even.
• Underwater (>20% loss): Number of holders and the total percentage of supply they hold who are currently down more than 20%.
• Avg hold duration: The average number of days holders have maintained their positions.

3. AI Advice (🤖 AI Advice)

Provides a synthesized risk/reward assessment:

• Rating: An overall sentiment rating (rating_em and rating_text).
• Core Risks: Specific identified dangers.
• Warnings: Signals to watch for.
• Positives: Favorable factors.
• Healthy chips: The percentage of 'healthy' supply (healthy_ratio) versus DEX-related supply excluded from evaluation.
• Exit Signals: Specific conditions that suggest it may be time to sell.

Use field combinations to determine the current lifecycle stage of a token to inform trading strategy.

Stage 1 — Early (New Born)

• Indicators: creation_timestamp < 1 hour ago, low hot_level, smart_degen_count = 0.
• Strategy: High risk. Wait for Stage 2 confirmation.

Stage 2 — Breakout

• Indicators: smart_degen_count ≥ 3 (and rising), surging volume (compare swaps_1h vs swaps_24h / 24), price_change_percent1h > 20%.
• Strategy: Strongest entry signal. Verify security before acting.

Stage 3 — Distribution

• Indicators: creator_token_status = creator_close, renowned_count buying, smart_degen_count plateauing/declining.
• Strategy: Late stage. Smart money may be exiting into retail/KOL demand.

Stage 4 — Decline

• Indicators: Declining volume/holders, high rat_trader_amount_rate, smart_degen_count declining.
• Strategy: Avoid new entries. Consider exiting existing positions.

The gmgn-wallet-score skill evaluates wallet addresses across three distinct dimensions to provide a comprehensive profile. It is important to distinguish between a trader's skill and the feasibility of following them.

• Track-record score (0–100): Measures actual profitability and risk management. It focuses on outcome distribution (e.g., how well a trader cuts losses) rather than just a simple win rate.
• Copy-tradeability score (0–100): Measures how easy it is for another user to capture the same edge. It penalizes strategies that are hard to replicate due to high latency, low margins, ultra-short hold times, or bot-tier frequency (e.g., sniping sub-$100k market caps).
• Backtest estimate: A projected return calculation: Raw Return - Entry-latency Price Drift - Round-trip Slippage - Gas.
• Dev-reputation score: Specifically for wallets that are primarily token creators (where created_token_count > 50% of traded_token_count). It evaluates the survival rate of their launches and security of recent tokens. Note that for Dev wallets, track-record and copy-tradeability scores are heavily discounted due to potential self-dealing.
• No trading history: If a wallet has no buy/sell transactions in the sampled period, the skill reports this explicitly instead of providing a score.