Dyuovo/china-power-market-agent

A production-ready multi-agent AI system design for power-market analysis, learning and operations, implemented on OpenClaw. Intended for practitioners building personal or team multi-agent deployments focused on electricity market reporting, price analysis, policy interpretation and structured learning workflows. Key elements: - Governance–execution separation: a governance agent handles task routing, system rules, memory archiving and backups; domain agents focus on market operations and learning/research. - Structured memory: daily memory → compression → semantic index → nightly reflection; local SQLite + hashed vectors, no external APIs required. - Task ledger and A2A protocol: SQLite task book with standardized message types (assign/ack/done/blocked), weekly/monthly retrospectives. - Three-stage learning pipeline: problem push, answer collection, graded/archived review written into DB. - Pre-submit self-review required for all external outputs. - Tech stack: OpenClaw, Lark (Feishu) messaging, SQLite storage, configurable local models, Python tooling, Docker deployment. Sensitive data, DB files, credentials and runtime configs are intentionally excluded from the repo.

AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.