Python SDK
pip install asahi
CLI Reference
Analyze from terminal
Interactive Lab
Test signals live
Introduction
AsahiLM is a meme token narrative intelligence engine for Solana. Language models read sequences — they weight context, measure coherence, track how meaning accumulates across positions. AsahiLM applies the same lens to on-chain meme tokens.
Every meme has a sequence: early signal, growing resonance, coherent narrative, viral escape velocity. Most traders read the chart. AsahiLM reads the structure beneath it — measuring whether a token's signals are coherent and directional, or just noise.
Installation
# Install from PyPI pip install asahi # No external dependencies. Pure Python 3.10+.
Quick Start
from asahi import classify, MemeTokenContext
ctx = MemeTokenContext(
token_mint="AsahiXyz...pump",
symbol="ASAHI",
price_usd=0.00082,
price_change_1h_pct=38.0,
price_change_24h_pct=180.0,
volume_1h=480_000,
volume_24h=1_200_000,
holder_count=4_200,
holder_count_prev=3_100,
market_cap_usd=560_000,
)
report = classify(ctx)
print(report.state) # NARRATIVE
print(report.narrative_score) # ~67.4
print(report.insight) # "Coherent narrative in motion..."
print(report.triggered_count) # 3
print(report.is_coherent) # TrueCore Concepts
Narrative Score
A 0-100 score representing overall narrative coherence. Higher scores indicate stronger, more directional market signals.
Phases
Four phases based on narrative score thresholds:
The Name
Asahi means morning sun in Japanese. The moment just before sunrise when light starts to differentiate signal from darkness. That's what this engine does: it reads the on-chain sequence and finds where the narrative is actually forming — before it becomes obvious.