Wiki Ingest

•
45
•

Ingests web pages, prompts, and notes into an interlinked markdown wiki with automated cross-referencing and lint checks.

Implementing Andrej Karpathy's Ingest primitive for LLM-driven knowledge management, this skill converts prompts, URLs, and synthesized markdown notes into atomic wiki pages called scraps. It searches the existing documentation graph, calibrates content depth based on graph familiarity heuristics, drafts concise markdown entries, and automatically converts unlinked plain-text mentions across existing documents into bidirectional wiki-links. Built-in linting ensures tag hygiene, prevents broken links, and maintains clean graph integrity without manual maintenance overhead.

주요 기능

0145 GitHub stars
02Familiarity heuristics that dynamically scale note length to prevent cognitive overload
03Automated bidirectional cross-linking using unlinked-mention lint discovery
04Automated graph sanity checking for broken links and source page alignments
05Source-agnostic ingestion from raw prompts, web URLs, and synthesized markdown
06Strict namespace validation separating page references from category tags

사용 사례

01Filing complex AI Q&A answers back into a persistent personal knowledge base
02Summarizing online research articles and seamlessly integrating them into a documentation graph
03Expanding project documentation by creating atomic concept scraps with automatic cross-linking