← Back to library

Clip bookmarks into a personal ask-me store

Miles Deutscher's second-brain card treats Grok as a personal database for bookmarks and articles instead of a heavy notes app. Our version is a clip-and-ask loop: you save, the agent indexes, and answers stay grounded in your store with citations.

  • OutcomeSave articles and bookmarks into a folder your agent indexes, then ask questions answered only from what you saved, with links back to the clips.
  • SetupMinutes
  • LevelStarter
  • CreditMiles Deutscher (@milesdeutscher)
View original
Works with
Grok BotChatGPT agentClaude

Published Oct 23, 2026Updated Oct 23, 2026AI dry run on sample input: passed Oct 23, 2026Source verified Oct 23, 2026

When should you use it?

  • You already save links and lose them when you need them later
  • You want answers from your own clips, not a fresh web search
  • You prefer a simple folder or notes store over a full wiki rollout

Not for: A shared company wiki or team knowledge base; Answering from memory or the open web when the clip is missing; Storing secrets, passwords or private credentials in the clip store

What do you need, and what do you get?

You need

  • A folder, notebook or bookmarks export the agent can read
  • How you will add new clips (browser bookmark, share sheet or paste)
  • The kinds of questions you usually ask (research, quotes, how-tos)

You get

  • An index of clips with title, URL, date saved and short gist
  • Answers that cite clip titles or URLs; "not in your store" when missing
  • Optional follow-up clips to save when a gap shows up often

How does it work, step by step?

  1. Pick the storeChoose one folder or notes notebook you control and point the agent at it.
  2. Clip as you goSave articles and bookmarks there with the URL and a one-line why-I-saved-this.
  3. Ask with groundingWhen you ask a question, require answers only from the store with citations.
  4. Fill gaps on purposeIf the agent says not in your store, decide whether to search and clip, or leave it.

The prompt

Our own prompt, inspired by Miles Deutscher (@milesdeutscher) and written to work in Grok Bot, ChatGPT agent and Claude. Values in lime are placeholders.

Fill in the blanks (optional; the copied prompt uses your values)

Prompt1 variable
You are my personal clip assistant for the store at {store_location}.
When I add a clip, record: title, URL, date, and a one-line gist from the page or my note.
When I ask a question:
- Answer only using clips in the store.
- Cite the clip title and URL for each claim.
- If the store does not cover it, reply exactly: Not in your store.
- Do not browse the live web for the answer unless I explicitly ask you to find something new to clip.
Never store passwords, API keys or private credentials. Keep this personal; do not treat it as a shared team wiki.

Inspired by Miles Deutscher (@milesdeutscher): X thread: 10 Grok Bot use cases (second brain, repurposer, subscription audit, personal X assistant). The prompt above is written by playbooks.bot in our own words; nothing from the original is quoted.

What can go wrong?

  • Do not let it invent a citation or quote that is not in a saved clip.
  • This is personal; do not treat it as a team wiki with shared edit rights.
  • Skip passwords, API keys and private customer data.
  • Re-index when you add a batch of clips so answers stay fresh.

Our take

Second-brain talk often balloons into a complex notes system. Deutscher's framing is useful when you keep it small: personal clips in, grounded answers out. The hard rule is citation or silence. Without that, the agent quietly blends your notes with generic web knowledge and you lose the point of saving anything. Distinct from company-docs RAG chatbots and shared team wikis, this stays personal and lightweight. You do not need a graph database or a shared edit workflow; a folder of links with gists is enough to make the ask loop useful. In our dry run we loaded eight fictional clips about home solar and battery sizing. Asked for a payback rule of thumb, the agent answered from two clips with URLs. Asked for local rebate amounts that were not saved, it replied not in your store and suggested one search query to clip later. No invented rebate figures. Use it when you already bookmark; do not start by promising a lifelong knowledge graph.

Where the time goes back: You stop re-finding the same articles and get answers with links back to what you already trusted enough to save.

FAQ

Is this a replacement for Obsidian?
No. It is a lighter clip-and-ask pattern. Keep Obsidian or any notes app if you like; the agent just answers from the folder you point at.
How is this different from a shared team wiki agents maintain?
shared-team-wiki-agents-maintain is for a group knowledge base in git. This playbook is a personal clip store you ask questions of.
What if the answer is not in my clips?
The agent should say not in your store. You can then search, save a new clip, and ask again.

Source and credit

Inspired by Miles Deutscher (@milesdeutscher). We describe the idea in our own words, link to the original, and wrote the prompt ourselves; nothing is quoted.

Inspired by
X thread: 10 Grok Bot use cases (second brain, repurposer, subscription audit, personal X assistant)
Creator
Miles Deutscher (@milesdeutscher)
Published on
X
Original date
Oct 23, 2026
License
Not openly licensed (the poster keeps all rights), so we quote nothing and link to the original
Originally for
Grok Bot
AI dry run
Oct 23, 2026: passed. Dry-run tested by our AI test agent on sample input. Grounded citations; silence on gaps; personal-store scope held.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails.
Verified
Oct 23, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.

Are you the creator? Claim or correct this page.

More in Career & Learning

All 37 Career & Learning playbooks →

Related playbooks

Career & LearningRoutine✓ AI dry run

Screen new job posts daily for level, pay and fit

Have an agent check new LinkedIn postings every morning, compare leveling and pay data, score fit against your resume and hand you a short list to judge.

Credit: malfist on Hacker News

Under an hour
Career & LearningRoutine✓ AI dry run

Copy homework due dates from a school portal to a calendar

Have an agent sign in to the school portal daily, pick out each assignment and its deadline, and put them on your kid's calendar along with an advance nudge.

Credit: fnordpiglet on Hacker News

Under an hour
Career & LearningWorkflow✓ AI dry run

Turn a contact sheet into job emails ready to review

Give an assistant a spreadsheet of hiring contacts and your resume, and get a personalized draft email per role waiting in Gmail for you to review and send.

Credit: Pratyush Pandey (@Pratyush2241)

Minutes
Career & LearningSkill✓ AI dry run

Turn any YouTube video into a short course you can take

Save a reusable skill that, whenever you share a YouTube link, asks what skill you hope to gain, pulls the transcript, plans a handful of focused lessons, builds a single HTML course file with quizzes, a final exam and a practice task tied to your goal, then offers follow-up quizzes on what you missed.

Credit: Michael Fenech (@Michael_Fenech_)

Under an hour