content-research

validates and invalidates every claim in a raw transcript using deep web research. produces a structured research file as input for the next pipeline steps.

inputs

• file: path to a transcript .md file (required)
• output goes to /Users/rami/Documents/life-os/notes-processing/{filename}/research.md where {filename} = the source file's name without extension

preprocessing (do this before anything else)

transcripts contain Obsidian syntax that must be cleaned before processing:

• strip YAML frontmatter (everything between --- and --- at the top)
• convert wiki-links: [[Note|display]] → display, [[Note]] → Note
• remove ## Related Notes section and everything after it
• work with the clean plain text from here on

steps