What problem does it solve? Turning long source materials like books and PDFs into searchable, concept-level notes is slow and error-prone when done manually, and a single-pass summary often misses niche material. This Skill deep-reads each raw source in a knowledge vault and produces richly tagged, per-concept summary notes that future questions can resolve directly. ## Core Features & Use Cases - Parallel subagent extraction: Dispatches one agent per source to read the entire document (including large PDFs page-by-page) and write one note per distinct concept into wiki/summaries/. - Schema-driven vault rules: Reads the vault's schema file first to respect its layers, frontmatter format, and hard rules such as no AI writes to wiki/atomic/ and no unconfident writeback. - Strict folder guardrails and failure recovery: Confines each subagent to its assigned source and output files, and resumes or relaunches agents after rate-limit failures without restarting completed work. - Use Case: After running /vault-inbox to create raw stubs for three new books, run this Skill to generate dozens of tagged summary notes per book, then update the vault index and log centrally. ## Quick Start Run /vault-summarize on all raw sources in my vault that do not yet have summary notes.
