amosblomqvist/learn

This issue highlights a critical interoperability failure: Mermaid diagrams generated by the AI learning system's `mermaid-maker` fail to parse in Obsidian due to version discrepancies and stricter parsing rules. The root cause lies in `mermaid-cli`'s lenient parsing versus Obsidian's older, more rigid implementation, particularly with unquoted labels containing reserved characters. This directly breaks the 'core learning flow' for users relying on Obsidian for knowledge management. The proposed fixes, including enforcing quoted labels and adding pre-publish validation, are essential for ensuring reliable output and user experience. For B2B SaaS, this underscores the necessity of robust compatibility testing across diverse client environments and the importance of adhering to stricter parsing standards or providing configurable output options. Failure to ensure seamless integration with popular third-party tools creates significant user friction and limits adoption.