200 Stars on GitHub. In 10 days.
Two weeks ago I published my first post about Semantic Anchors -- the idea that five precise words like "TDD, London School" can outperform a 200-line system prompt because they activate what the model already knows.
The response blew me away. 200+ stars, mass of contributions, and discussions that pushed the concept further than I could have on my own.
What started as a personal observation has become a community project: 52 curated anchors across software architecture, testing, design patterns, and development methodologies. Each one tested against the quality criteria (precise, rich, consistent, attributable).
A few highlights from the contributions:
- New anchors I hadn't considered - Quality improvements to existing entries - Thoughtful debates about what qualifies as an anchor vs. a buzzword
Thank you to everyone who starred, contributed, or challenged the idea. Open source lives from exactly this.
If you find Semantic Anchors useful in your daily work with LLMs, a star on GitHub helps the project reach more people: https://github.com/LLM-Coding/Semantic-Anchors
What anchor would you add?