1 post
What 150K LLM citations tell builders about prompt-time grounding, eval coverage, and the source biases their agents inherit by default.