Show HN: d2-mcp – compile and render D2 diagrams for multimodal LLMs (github.com)

Would love to hear how you see d2-mcp fitting into broader AI-assisted workflows — especially in things like technical writing or architecture docs.