Create Bookmark

Save a URL as a dated bookmark file in the Brain with structured frontmatter and a human blurb or AI-generated summary.

Overview

Bookmarks are stored as individual markdown files in Bookmarks/YYYY-MM-DD/ with sequential numbering, following the same pattern as Daily Projects. Each bookmark has frontmatter (uid, type, url, title, source, tags) and a markdown body containing the user's blurb or an AI summary of the linked content.

When to Use

Use this skill when the user:

• Says "bookmark this", "save this link", "bookmark that URL"
• Shares a URL and asks you to save or remember it
• Asks to import bookmarks from a file or list
• References a link they want to come back to later

File Format

Bookmarks/
  2026-02-13/
    01 using an engineering notebook.md
    02 roblox sentinel risk detection.md
  2026-02-14/
    01 spiral funnel adr.md

Each file looks like:

---
uid: 3medhoffc22xy
type: bookmark
created: 2026-02-13T00:00:00Z
url: https://ntietz.com/blog/engineering-notebook/
title: "Using an Engineering Notebook"
source: hacker-news
tags: [pkm, architecture]
---
Nicole's take on engineering notebooks as thinking tools, not reference material.
The key insight is that she rarely reads her notes back — "the writing is the
work, in a lot of ways, since it's the thinking."

Frontmatter Fields

Body

The markdown body contains either:

• User's blurb — if they provide context about why they're saving it
• AI summary — if no blurb is given, fetch the URL content and write a 1-3 sentence summary explaining what the page is about and why it's interesting

Workflow

Step 0: Sync Brain from remote

Before doing anything else, pull the latest changes from the remote to ensure you're working with the most current Brain content:

cd ~/Brain && git pull --rebase --autostash -X theirs

If rebase fails, fallback to merge:

cd ~/Brain && git rebase --abort && git pull --no-rebase -X theirs --no-edit

Step 1: Extract the URL

Get the URL from the user's message. If they share multiple URLs, create one bookmark per URL.

Step 2: Get context from the user

Check if the user provided:

• A blurb (their own description of why this is interesting)
• A source (where they found it: hacker-news, slack, colleague, twitter, etc.)
• Tags from the tag vocabulary

If any of these aren't provided, that's fine — they're all optional except the blurb/summary.

Step 3: Fetch the page title

Use the WebFetch tool to load the URL and extract the page title. If the fetch fails, derive a title from the URL path.

Step 4: Generate AI summary (if no blurb)

If the user didn't provide a blurb, use the fetched page content to write a 1-3 sentence summary. Write it in first person as if explaining why the user saved it. Be concise and focus on what makes it interesting or useful.

Step 5: Create the bookmark file

Create the file at Bookmarks/YYYY-MM-DD/NN title.md where:

• YYYY-MM-DD is today's date
• NN is the next sequential number in that date folder
• title is the slugified page title (lowercase, spaces preserved, unsafe chars stripped)

Use the generateTid() function for the UID, or generate a TID manually (13-char base32-sortable timestamp).

Step 6: Confirm to the user

Tell the user the bookmark was created and show the path. If they want to add tags or edit the blurb, they can do so directly.

Examples

User provides blurb and source

User: bookmark this https://ntietz.com/blog/engineering-notebook/ — found it on HN, great take on using notebooks as thinking tools not reference material

Agent creates:
  Bookmarks/2026-02-13/01 using an engineering notebook.md
  with source: hacker-news
  with user's blurb as body

User provides just a URL

User: save this link https://corp.roblox.com/newsroom/2025/08/open-sourcing-roblox-sentinel-preemptive-risk-detection

Agent:
  1. Fetches the page
  2. Extracts title: "Open Sourcing Roblox Sentinel: Preemptive Risk Detection"
  3. Generates AI summary from page content
  4. Creates: Bookmarks/2026-02-13/01 open sourcing roblox sentinel preemptive risk detection.md

Multiple bookmarks at once

User: bookmark these:
- https://example.com/article1
- https://example.com/article2
- https://example.com/article3

Agent creates three separate files in Bookmarks/YYYY-MM-DD/, numbered 01, 02, 03.

Querying Bookmarks

Users may ask to find bookmarks. When they do:

• Search Bookmarks/ directory by filename, tags, or content
• Use grep/glob to find bookmarks matching keywords
• Report results with the URL, title, and blurb

Important Notes

• The Bookmarks/ directory is in the Brain root (~/Brain/Bookmarks/)
• One file per bookmark, organized into date folders
• Sequential numbering within each date folder (same pattern as Daily Projects)
• Tags should come from the Brain's tag vocabulary when possible
• The source field uses lowercase hyphenated names (e.g. hacker-news, not Hacker News)
• Titles are quoted in frontmatter to handle special characters
• If the URL is unreachable, still create the bookmark with whatever title/info is available