My Web Clipper Prompt (LYT)#

Original | Raw

Author: Nick Milo (Linking Your Thinking)

Summary#

Setup guide for the Obsidian Web Clipper’s Interpreter feature. Uses three AI-powered prompts (via {{" "}} syntax) to generate structured layers above any clipped article:

  1. Summary — most interesting aspect as a single declarative statement
  2. Headlines — bold one-line insights with 1-2 supporting sentences
  3. Things — bulleted list of people, places, numbers, things, and concepts

Model-agnostic — works with Claude, GPT, Gemini, DeepSeek, or local models via Ollama.

Significance#

A lightweight, zero-code approach to knowledge capture that complements the llm-wiki-pattern:

  • Web Clipper: capture-time processing, single-article scope, browser-native
  • LLM Wiki: ingest-time processing, cross-source synthesis, vault-native

Both share the principle: AI processes, human curates. The three-layer template (summary → headlines → entities) mirrors this wiki’s own source page structure.

Connections#

  • Complements notebooklm-notes-guide — both are “AI-enhanced capture” tools with different scopes
  • Validates llm-wiki-pattern — same capture→process→organize flow, lighter weight
  • Connects to context-management — the interpreter prompt is a micro-context-engineering exercise

See Also#

Backward Llm Wiki Karpathy Mem0 Memory Management Forward