voice-memo-triage

Overview

The aggregate voice memo file collects dictated memos newest first, each under a ## <weekday>, <Month> <D>, <YYYY> at <H>:<MM> <AM|PM> header.

Some of those memos are questions or things that need to be discovered, and those belong in their own research notes.

The rest are reminders, shopping notes, transcripts of conversations and ambient chatter, and those stay in the aggregate.

This skill takes a window of entries, splits them into those two piles, researches the first pile with a workflow, and removes exactly those entries from the aggregate.

Nothing is ever discarded — every removed entry survives verbatim inside its research note's ## Original request section, and the removal is verified by byte accounting before the commit.

Parameters

• aggregate — absolute path to the aggregate file to drain; default /Users/mtm/Documents/Obsidian Vault/claude aggregate voice memos.md
• scope — which entries to consider, either a count of the newest entries or one transcription run; default a count
• count — how many of the newest entries to consider under a count scope; default 10
• scratch directory — $CLAUDE_JOB_DIR/tmp when that variable is set, otherwise mktemp -d