NotebookLM Prompts Collection

1What it does

A straightforward prompt bank: dozens of NotebookLM-ready prompts grouped by the task they perform — study aids (quiz generation, flashcard seeding, explanation prompts), research flows (literature synthesis, claim extraction, source comparison), writing support (outlines, summaries at a target length, tone shifts) and analysis patterns (compare-and-contrast across sources, gap finding). The collection's value is task coverage rather than depth on any one pattern: where the Prompt Lab experiments, this repo indexes — each entry is a copy-paste block with a short intent note, which makes it faster to scan when you know what job you need done. Prompts are written to be source-grounded — they assume a loaded notebook and lean on NotebookLM's citation behavior rather than fighting it.

2When to use it

Reach for it as a lookup table: you know the task (make a quiz, synthesize three PDFs, draft a briefing) and want a prompt that already encodes the source-grounding phrasing. It suits students and researchers who use NotebookLM episodically — grab, adapt, move on — and teams seeding an internal prompt doc, since the task-grouped structure ports cleanly into a wiki. Skip it if you want prompts with testing notes and iteration history — this is an index, not a lab notebook; the depth lives in the Prompt Lab repo. Also less useful once you've internalized the source- grounding patterns — at that point your own phrasing will beat the generic one.

3How to get started

Clone or browse the repository, find the task group matching your job, and copy the prompt into a notebook with sources already loaded. Adapt the nouns — subject, audience, output format — but keep the grounding clauses ("from the sources", "with citations") intact. For a first pass, test the prompt on a notebook where you know the right answer; a source-grounded prompt that hallucinates usually means a phrasing problem or an unparsed source, not a model limitation. No dependencies, no build step — it's markdown you read and copy.

4Configuration

The tunable surface is the prompt tail: output length, format (bullets vs prose vs table), audience level, and source scope. Some prompts name a source count ("across all sources" vs "in the selected source") — NotebookLM honors source selection in the UI, so the prompt wording should match what you actually selected in the sources panel. When adapting for a team, standardize the grounding clause once and vary the task clause per use — mixing grounding phrasings across prompts is the most common way to get inconsistent citation behavior.

5Things to know

• Community-maintained — coverage reflects what the contributors needed; niche analysis tasks may not have a ready pattern.
• Task-grouped rather than difficulty-grouped — a "beginner" prompt can still produce expert output if the sources are dense.
• Prompt quality is prompt-dependent — these are starting points; the real skill is editing the task clause, not finding a perfect ready-made.
• MIT-licensed; the repo's flat structure makes it trivial to vendor into your own docs or prompt manager.
• Because prompts are grouped by task, the fastest way in is the README's section list — jumping straight to your job beats reading the bank linearly.
• Study prompts are the deepest section — quiz and flashcard seeds are the most-tested patterns, which tracks with the collection's student-heavy contributor base.
• Combine with a flashcard export tool for the full loop — the prompts generate the Q&A text; a separate tool (see the flashcards hub) turns it into an Anki deck.
• Adaptation is the design intent — each prompt leaves obvious slots (subject, audience, format) so the pattern survives the edit that makes it yours.
• The bank stays useful as a seed set — clone once, prune to your top ten, and keep the rest as reference for the jobs you haven't hit yet.

6Related on the map

References

• Repository
• NotebookLM help center