I've open-sourced a catalog with 21+ Semantic Anchors covering Software Architecture, Testing, Requirements, Decision-Making, and Documentation. Available on GitHub.
What Are Semantic Anchors?
Semantic Anchors are established terms that activate coherent response patterns in LLMs. Instead of explaining what you want in detail, you use a term the LLM already "knows" from its training data.
The Catalog Categories
🏗️ Software Architecture
- C4 Model - Context, Containers, Components, Code diagrams
- arc42 - Architecture documentation template
- ADR - Architecture Decision Records
- Domain-Driven Design - Strategic and tactical patterns
- Ports & Adapters - Hexagonal architecture
🧪 Testing
- TDD London School - Outside-in, mock-heavy approach
- TDD Detroit School - Inside-out, classical approach
- Given-When-Then - BDD scenario format
- Arrange-Act-Assert - Test structure pattern
- Test Pyramid - Unit, Integration, E2E balance
📋 Requirements
- User Story - As a [role], I want [feature], so that [benefit]
- Job Story - When [situation], I want [motivation], so I can [outcome]
- INVEST Criteria - Independent, Negotiable, Valuable, Estimable, Small, Testable
- Acceptance Criteria - Conditions of satisfaction
🎯 Decision Making
- SWOT Analysis - Strengths, Weaknesses, Opportunities, Threats
- Pros/Cons Matrix - Weighted comparison
- 5 Whys - Root cause analysis
- Cost-Benefit Analysis - Economic evaluation
📝 Documentation
- Docs-as-Code - Documentation in version control
- Diátaxis Framework - Tutorials, How-tos, Reference, Explanation
- Pyramid Principle - BLUF, MECE, SCQ
How to Use
Instead of:
"Write tests that first set up the context, then perform an action, then verify the result. Make sure each test is independent..."
Just say:
"Write tests using Arrange-Act-Assert pattern"
Open Source
The full catalog is available on GitHub. Contributions welcome!