Overview
Seshat Documentation
Technical deep-dives into each of Seshat's four core intelligence systems. Written for developers and researchers who want to understand the implementation, methodology, and data pipelines behind the scores.
Fee Scanner
Multicall batch fee reads, V0–V4 ABI fallback, Bitquery event indexing, first-ever claim detection.
Developer Scoring
FrontrunPro wallet analysis, 4-dimension composite score, hidden gem detection algorithm.
AI Research Loop
Agentic Seshat pipeline, LARP detection methodology, GitHub analysis, novelty scoring.
Self-Learning Model
Ridge regression, outcome labeling, weight update cadence, R² tracking and validation.
Architecture overview
Seshat is a Cloudflare Workers TypeScript application with zero server infrastructure. All compute runs on the edge; persistence is Cloudflare D1 (SQLite) and KV; storage is R2. Scheduled cron triggers drive the three main pipelines.