Fabric: The Best AI Prompts for Obsidian

Fabric AI patterns are pre-made prompt templates—about 140 of them—that can be applied inside Obsidian to turn raw notes into structured outputs like summaries, organized ideas, claim verification, and marketing deliverables. The core value is speed and consistency: instead of crafting prompts from scratch, users select a pattern and apply it to existing notes, effectively extending what their notes can produce.

The workflow starts with Fabric’s definition as an open-source framework for augmenting humans with AI. While the project includes a broader philosophy and documentation, the practical focus here is the “patterns” library—each pattern is a ready-to-run prompt with instructions for what to extract and how to format the result. One example, “create five sentence summary,” includes a specific formatting and extraction recipe, even down to a playful persona line (“all knowing AI with a 476 IQ”), followed by step-by-step guidance for generating the five-sentence output.

To make the patterns usable, the setup integrates them into the Fonzer 2000 plugin for Obsidian (the transcript’s integration point). Because the library is large, the creator first generates an overview of what each pattern does by pasting the patterns into the Fonzer 2000 AI chat and asking for a cleaned list (e.g., “extract only names”). That produces a numbered catalog users can scan quickly, then selectively keep.

From there, the process becomes “highlight and apply.” Users create a dedicated “fabric” folder inside Obsidian, paste the downloaded patterns into it, and enable the Fabric formatting option in Fonzer 2000 settings. Once enabled, patterns appear in the assistant sidebar and can be applied directly to open notes.

Several concrete examples show how the patterns behave in practice. For social content, a “tweet” format pattern can generate multiple tweet drafts from existing Twitter post drafts, including emojis. For business ideation, “create Horoi offer” analyzes a “brain dump” of agency ideas and produces an analysis plus three offers derived from the source material. For knowledge organization, “capture thinkers work” takes book notes (the transcript references Nicholas Taleb’s Antifragility) and produces a structured one-line explanation of schools of thought, along with the thinkers involved.

The transcript also emphasizes control and safety: users can undo changes (Command Z), and the feature is described as experimental—meaning patterns can be deleted, edited, or customized, and the development direction depends on user feedback. Overall, Fabric’s pre-made prompt patterns plus Fonzer 2000’s in-Obsidian integration aim to make note transformation routine rather than a bespoke prompt-writing task.