Ginko
GINKO
[DOCUMENTATION]

Ginko Docs

01

Overview

銀子Ginko is a multi-agent LLM system for detecting silent accumulation phases in Solana tokens. 銀子 (Ginko) — Silver child. The Ginko tree waits in silence for centuries. Then blooms all at once. The best Solana opportunities don't announce themselves. They sit in silence while smart money quietly builds positions. The community is small. The charts are flat. Nobody's talking about it. Then — discovery. The narrative breaks through. Price moves violently. By the time retail notices, the bloom has happened. Ginko detects the pre-bloom phase.

02

Four Agents

Ginko uses four independent LLM agents: DormancyAnalyzer — Measures silence depth, quiet day count, and pattern classification. How long and how deeply has this token been dormant? HolderAnalyzer — Analyzes veteran ratio, diamond hands, and organic growth detection. Who is holding and are they real? AccumulationAgent — Detects smart wallet patterns, buy/sell ratios, and conviction scores. Is smart money quietly entering? DiscoveryAgent — Estimates community readiness, trigger signals, viral coefficient. How close is market discovery? Each agent runs independently. Scores are weighted and combined into a single Bloom Index (0-100).

03

Installation

[terminal]
pip install ginko

# Or from source
git clone https://github.com/theGinkodev/-Ginko.git
cd -Ginko
pip install -r requirements.txt
cp .env.example .env

# With Docker
docker compose up
04

Configuration

Without an API key, all agents run in heuristic mode (no LLM calls). Full test suite passes without a key.

[terminal]
ANTHROPIC_API_KEY=your_key_here
GINKO_MODEL=claude-haiku-4-5-20251001
GINKO_CACHE_TTL=300
05

Usage

[terminal]
from ginko import GinkoEngine, TokenSnapshot
import time

engine = GinkoEngine()

snapshots = [
    TokenSnapshot(
        token_ca="YourCA...",
        token_name="TOKEN",
        timestamp=time.time() - 86400,
        holder_count=450,
        holder_growth_24h=1.8,
        top10_concentration=0.28,
        lp_locked=True,
        lp_lock_days=180,
        volume_24h=18000,
        market_cap=120000,
        discord_members=650,
        discord_activity=0.12,
        twitter_mentions=45,
        twitter_sentiment=0.2,
        developer_activity=3,
        days_since_launch=38,
        previous_peak_mc=350000,
        current_vs_peak=0.34,
    )
]

result = engine.analyze("YourCA...", "TOKEN", snapshots)
print(f"Bloom Index: {result.bloom_index}/100")
print(f"Phase: {result.bloom_phase.value}")
print(f"Action: {result.recommended_action}")
print(f"Est. bloom days: {result.estimated_bloom_days}")
06

Bloom Index Weights

Accumulation (35%) — Smart money entry patterns Dormancy (25%) — Length and depth of silence Discovery (25%) — Proximity to market discovery Holder Quality (15%) — Veteran ratio, organic growth

07

Development

[terminal]
git clone https://github.com/theGinkodev/-Ginko.git
cd -Ginko
pip install -r requirements.txt
pytest tests/ -v   # 78 tests, all pass without API key