AgentSkillsCN

pokedao-architecture

PokeDAO 项目架构与结构指南。适用场景如下： (1) 开发 PokeDAO 的任意功能或修复 Bug； (2) 需要深入理解单体仓库的组织结构； (3) 与 Telegram 机器人、机器学习流水线或信号系统协同工作； (4) 对于“应在何处新增功能”这一问题存疑时。 核心组件：Telegram 机器人（Grammy）、机器学习信号、Prisma 数据库、Redis 缓存。

SKILL.md
--- frontmatter
name: pokedao-architecture
description: |
  PokeDAO project architecture and structure guide. Use when:
  (1) Working on any PokeDAO feature or bug fix
  (2) Need to understand the monorepo structure
  (3) Working with the Telegram bot, ML pipeline, or signal system
  (4) Questions about where to add new functionality
  Key components: Telegram bot (Grammy), ML signals, Prisma DB, Redis cache
author: Claude Code
version: 1.0.0
date: 2025-01-19

PokeDAO Project Architecture

Problem

PokeDAO is a complex monorepo with multiple apps and packages. Understanding where code lives and how components interact is essential for effective development.

Context / Trigger Conditions

  • •Working on any PokeDAO feature
  • •Adding new bot commands or handlers
  • •Modifying the ML/signal pipeline
  • •Debugging cross-package issues
  • •Understanding data flow

Solution

Project Structure

code
pokedao/
├── apps/
│   ├── agent/          # Signal processing agent
│   │   └── src/tick.ts # Main tick loop for signals
│   └── mew1a/          # ML/vLLM deployment
├── bot/                # Telegram bot (Grammy)
│   └── src/
│       ├── index.ts           # Bot entry point
│       ├── commands/          # Bot commands (/start, /wallet, /alerts, etc.)
│       ├── callbacks/         # Callback query handlers
│       ├── alerts/            # Alert formatting and sending
│       ├── middleware/        # Auth, rate limiting
│       └── lib/               # Shared utilities (config, logger, prisma)
├── ml/                 # ML system
│   └── src/alertSystem.ts    # Alert generation from signals
├── packages/           # Shared packages
│   ├── core/           # Core utilities
│   ├── storage/        # Database/storage layer
│   └── analysis/       # Analysis tools
└── prisma/             # Database schema

Key Technologies

  • •Bot Framework: Grammy (Telegram)
  • •Database: Prisma + PostgreSQL
  • •Cache: Redis (ioredis)
  • •Logging: Pino
  • •Validation: Zod
  • •Runtime: Node.js with TypeScript

Adding New Bot Commands

  1. •Create command file in bot/src/commands/[name].ts
  2. •Export handler function
  3. •Register in bot/src/index.ts with bot.command()

Adding New Alert Types

  1. •Define alert structure in ml/src/alertSystem.ts
  2. •Add formatter in bot/src/alerts/formatter.ts
  3. •Update sender logic in bot/src/alerts/sender.ts

Verification

  • •Run pnpm typecheck from root to verify types
  • •Run pnpm dev in bot/ to test bot locally
  • •Check logs with pino-pretty for debugging

Notes

  • •This is a pnpm workspace monorepo
  • •Use workspace dependencies: @pokedao/core, @pokedao/storage, @pokedao/analysis
  • •Environment variables defined in .env.example
  • •Deployed via render.yaml configuration