AI Coding Principles A collection of Claude Code skills for coding discipline and system design knowledge. 中文文档 Skills ai coding discipline Mandatory rules loaded during all code writing tasks that prevent common AI coding anti patterns. Rule Summary 1 No Silent Fallbacks Don't use ?? / to mask values that should never be missing 2 No Catch All try/catch Business logic lets errors propagate; catch only at API boundaries 3 Tests Must Fail When Code Breaks Verify specific outcomes, not just existence 4 No Hardcoded Lookup Tables Implement real logic, not test case fitting shims 5 Red Green Testing (TDD) Write failing test first, then fix 6 Don't Remove Debug Logs During Fix Logs stay until human confirms the fix works ddia principles Distilled reference guide based on Martin Kleppmann's Designing Data Intensive Applications . Loaded when designing database schemas, choosing storage engines, implementing replication/partitioning, handling distributed transactions, or building batch/stream processing pipelines. Part Topics I: Foundations Reliability, Scalability, Maintainability; Data Models & Query Languages; Storage & Retrieval (B tree vs LSM tree, OLTP vs OLAP); Encoding & Evolution II: Distributed Data Replication (single/multi leader, leaderless); Partitioning (key range, hash, compound); Transactions & Isolation Levels; Distributed System Challenges; Consistency & Consensus III: Derived Data Batch Processing (MapReduce, Spark, Flink); Stream Processing (Kafka, CDC, Event Sourcing); Data Integration Patterns Installation [安装/使用命令见下方] License MIT