What problem does it solve? Large document collections are hard to navigate: researchers cannot quickly find which files discuss which concepts, and AI answers lose traceability to sources. This Skill batch-reads raw files, extracts grounded summaries, key passages, and concept signals, then builds Obsidian-linked navigation maps so every claim traces back to a file and line. ## Core Features & Use Cases - Batch Extraction: Reads raw files in batches and writes extraction packets with summaries, key passages with line references, concept signals, and Obsidian tags to agent_reports/extraction_{batch_id}.md. - Navigation Map Writing: Creates and enriches hub, group, and theme maps in maps/ using Obsidian wikilinks and hierarchical tags for graph-view filtering. - Idempotency & Safety: Skips already-processed batches, marks unreadable files, and enforces token limits for free-tier models. - Use Case: A social scientist with 200 interview transcripts runs the mapper to produce extraction packets and a corpus overview map linking every transcript to its themes and concepts. ## Quick Start Ask the agent to run a map_extract batch over the files in raw/ and write the extraction packet to agent_reports.
