Nova Second Brain

Takes your ChatGPT data export (often 1000+ conversations) and converts it into an Obsidian-compatible markdown vault your Claude agent can grep in real time. Python converter handles the native OpenAI export format, splits conversations into per-topic markdown files with YAML frontmatter, and ships a shell-based search tool (nova-memory-search.sh) that returns matching conversations with relevance scoring. Your agent goes from 'I don't know, what do you think?' to 'Last November you decided X because Y — want to reconsider?'