Skip to content

bug: conversation-parser no_match on pages with 78% pattern match rate #1533

@garrytan-agents

Description

@garrytan-agents

Problem

conversation-parser scan reports no_match on pages where 77.8% of lines match the imessage-slack pattern.

Reproduction

Page has 226 lines, 175 of which are valid imessage-slack format:

**Garry Tan** (2026-01-29 12:00 PM): Hey man, how's it going?
**Participant 2** (2026-01-29 12:00 PM): Hey Gary, how you doing man?

A minimal page with ONLY these lines → regex_match, matched_pattern_id: imessage-slack, 4 messages ✅

The same lines inside a full meeting page (with frontmatter, ## Summary, ## Transcript header, a blockquote reference) → no_match, unmatched_lines: 200

Expected

77.8% match rate should exceed any reasonable confidence threshold. The parser reports 200 unmatched lines when there are only 51 non-transcript lines.

Likely cause

The parser may be counting lines from compiled_truth + timeline (the full DB representation) rather than just compiled_truth. Or the unmatched_lines count includes frontmatter/YAML lines that shouldn't be counted.

Environment

  • gbrain 0.41.17.0
  • 36 Circleback meeting pages reformatted to imessage-slack pattern
  • All pages have ## Transcript section with transcript lines after it

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions