{"id":29522,"date":"2026-05-06T14:05:15","date_gmt":"2026-05-06T14:05:15","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/29522\/"},"modified":"2026-05-06T14:05:15","modified_gmt":"2026-05-06T14:05:15","slug":"how-to-use-claude-with-obsidian-to-quickly-build-a-second-brain-prompts-included","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/29522\/","title":{"rendered":"How to use Claude with Obsidian to quickly build a second brain (prompts included)"},"content":{"rendered":"

Ever tried building a second brain, only to abandon it three weeks in? I have\u2014more times than I\u2019d like to admit. Capturing ideas<\/a> is the easy part; organizing, tagging, and maintaining everything is where it falls apart. But I\u2019ve found a cheat code: using Obsidian to capture ideas while Claude handles the organization. Here\u2019s how my system works\u2014and the exact prompts I use.<\/p>\n

Here\u2019s how the workflow comes together<\/p>\n

Obsidian stores, Claude organizes<\/p>\n

When I say \u201csecond brain,\u201d I\u2019m referring to Tiago Forte\u2019s concept<\/a>\u2014a personal knowledge management (PKM) system designed to offload remembering so you can focus on thinking. It follows a four-step process known as CODE:<\/p>\n

Capture: Save what resonates<\/p>\n

Organize: Structure it for retrieval<\/p>\n

Distill: Reduce it to its essence<\/p>\n

Express: Turn it into output<\/p>\n

It\u2019s a powerful system for offloading what you learn and structuring it for later use. But building and maintaining a second brain takes work: organizing notes, tagging them, and revisiting them to distill insights. That\u2019s why many people start strong but burn out shortly after.<\/p>\n

Claude changes that. With the Claude desktop app, you have access to Cowork mode<\/a>. This gives the AI access to your local files, allowing it to read, organize, and even modify your notes.<\/p>\n

That still leaves one problem: raw folders full of text files are hard to navigate. This is where Obsidian comes in. You can turn any folder into an Obsidian vault, giving you a clean, structured interface to browse your notes. Combined, this creates a streamlined second brain: Obsidian acts as your local, private knowledge base, while Claude works as an intelligent layer that organizes and maintains it.<\/p>\n

In the following sections, I\u2019ll break down exactly how this works in practice.<\/p>\n

Using Obsidian to \u2018capture\u2019 ideas<\/p>\n

This is the easy part<\/p>\n

\"Obsidian<\/p>\n

The first step of any second brain is capturing information before it vanishes\u2014and Obsidian handles this well across different contexts. For your own thoughts, reflections, and quick logs, Obsidian\u2019s built-in daily notes feature<\/a> works as a frictionless inbox. Technically, it\u2019s just a new note for each day, but having a consistent place to capture ideas means you\u2019re never deciding where something should go.<\/p>\n

Obsidian can also help you store ideas you encounter throughout the day. If you\u2019re on desktop, you can use the Obsidian Web Clipper<\/a> to save full pages or selected highlights directly into your vault. Alternatively, you can download content and save it straight into your vault. On mobile, the Obsidian app integrates with the system share sheet on both iPhone and Android. As such, you can send text and links directly to your notes from almost anywhere.<\/p>\n


\n \"Obsidian<\/p>\n

<\/a><\/p>\n

Related<\/p>\n

\t\t
\n\t\t\tHow I save any web page to my Obsidian vault (and actually read it later)
\n\t\t<\/a><\/p>\n

Obsidian replaced every read-it-later app I tried. Here’s my simple clipping workflow so articles stay clean, searchable, and useful.<\/p>\n

The capture process doesn\u2019t have to be entirely manual either. With Claude in Cowork mode, you can have the AI research a topic, browse sources, compile notes, and save the results directly into your Obsidian vault as structured Markdown files. Finally, to ensure the Obsidian vault on your PC is synced with the one on your phone, you can use Syncthing<\/a>. It\u2019s free, open-source, and frankly a must-have if you use local-first tools like Obsidian.<\/p>\n


\n \"A<\/p>\n

<\/a><\/p>\n

Related<\/p>\n

\t\t
\n\t\t\tThis free, open-source tool solves my biggest problem with local-first apps
\n\t\t<\/a><\/p>\n

Syncthing fixes the biggest problem with local-first apps. Notes, passwords, and documents stay private without being trapped on one machine.<\/p>\n

Using Claude to \u2018organize\u2019 your captured ideas<\/p>\n

One prompt to automate the filing you’ve been putting off<\/p>\n

Once ideas are captured, they need a structured home. Now, the standard approach to organizing your notes for a second brain is Tiago Forte\u2019s PARA method<\/a>\u2014Projects, Areas, Resources, and Archives\u2014four categories that cover most use cases. In practice, however, maintaining PARA often means manually sorting and moving files, which quickly turns into a tedious chore.<\/p>\n

This is where Claude comes in. At the end of the day, I just open Claude, switch to Cowork mode, and give it a brief update on my current priorities. With access to my vault, it scans my notes and inbox, determines where each item belongs, and organizes everything accordingly.<\/p>\n

Here\u2019s the prompt I use:<\/p>\n

You are my personal knowledge librarian for my Obsidian vault. Your job is to organize loose notes into the PARA structure (Projects, Areas, Resources,Archives) while keeping my tag vocabulary consistent over time. Work in four phases and wait for me between phases where noted.<\/p>\n

PHASE 1 \u2014 Get the vault path<\/p>\n

Ask me for the absolute path to my Obsidian vault. Wait for my answer.<\/p>\n

PHASE 2 \u2014 Scan and analyze silently
\nRead every note in:
\n– The “Daily Notes” folder
\n– Any loose .md files outside the PARA folders (Projects\/, Areas\/, Resources\/, Archives\/)Ignore anything already inside a PARA folder \u2014 it’s already organized.<\/p>\n

Also read these two meta files at the vault root:<\/p>\n

– \/_claude-index.md \u2014 the map of previously organized notes. Read it to understand the entry format in use.
\n– \/_claude-tags.md \u2014 my approved tag vocabulary. Only these tags may be applied to notes.
\nIf either file doesn’t exist, note that \u2014 you’ll bootstrap it in Phase 3 or Phase 4.<\/p>\n

As you read, create a mental model of the notes by theme and form preliminary PARA judgments.<\/p>\n

PARA definitions:
\n– Projects: active work with a deadline or defined outcome
\n– Areas: ongoing responsibilities with no end date (health, finances, home)
\n– Resources: topics of interest saved for future reference
\n– Archives: inactive items from the above three<\/p>\n

The same topic can land in different buckets depending on my life context \u2014 a note on “running shoes” is a Project if I’m shopping for race-day shoes,an Area if running is part of my lifestyle, and a Resource if it’s reference. You don’t know my context yet \u2014 that’s what Phase 3 is for.<\/p>\n

Do NOT move any files, modify any tags, or ask any questions yet. Just read
\nand analyze.<\/p>\n

PHASE 3 \u2014 Ask informed, leading questions<\/p>\n

Show me a brief summary of what you found (counts grouped by theme, not PARA bucket yet). Then ask me a batched questionnaire \u2014 all questions in a single message so I can answer in one reply. Three types of questions:<\/p>\n

(a) CONTEXT questions, derived from the themes you found. Be specific and leading \u2014 don’t ask “what are you working on?” in the abstract. Ask about what’s actually in the notes. Examples:<\/p>\n

“I found 6 notes about marathon training from the past two weeks. Are you:
\n a) Actively training for a specific race with a date
\n b) Running as an ongoing part of your lifestyle
\n c) Just saving reference material for later”<\/p>\n

“There are 4 notes about a kitchen renovation. Is this:
\n a) A current project you’re managing
\n b) A finished project (archive it)
\n c) Research for a future renovation”<\/p>\n

(b) DISAMBIGUATION questions for individual notes that don’t fit cleanly into a theme or have no clear PARA home:<\/p>\n

“Note: ‘notion-vs-obsidian-comparison.md’
\n Which bucket?
\n a) Projects \u2014 evaluating tools for a specific decision
\n b) Resources \u2014 general reference
\n c) Archives \u2014 no longer relevant
\n d) Leave it alone”<\/p>\n

(c) TAG VOCABULARY questions:
\n– If _claude-tags.md doesn’t exist yet: propose a starter vocabulary of 5-8 tags based on the themes you found, each with a one-line description of what it covers. Ask me to approve, edit, or reject.
\n– If it exists but some notes don’t fit any approved tag: list the proposed additions (e.g., “I’d like to add #finance for 3 notes about budgeting \u2014 approve?”). Never add a tag to the vocabulary without asking.
\n– If everything fits the existing vocabulary: skip this question.<\/p>\n

Batch all (a), (b), and (c) questions in one message. Wait for my answers.<\/p>\n

PHASE 4 \u2014 Execute
\nOnce I’ve answered:<\/p>\n

1. If I approved new tags (or a starter vocabulary), update _claude-tags.md. If the file didn’t exist, create it with a short header explaining its purpose as my approved tag vocabulary, then list each tag with its one-line description.<\/p>\n

2. If _claude-index.md doesn’t exist, create it with a short header explaining its purpose as a map of the vault for future sessions.<\/p>\n

3. For each note being organized:
\n a. Add 2-3 Obsidian tags to the frontmatter, using ONLY tags from the approved vocabulary in _claude-tags.md. If a note doesn’t fit any approved tag well, leave it untagged rather than forcing a bad match or inventing a new tag on the fly.
\n b. Move the file into the correct PARA folder.
\n c. Append an entry to _claude-index.md matching the existing format, roughly:
\n– [[note-filename]] \u2014 one-line abstract. Tags: #tag1 #tag2<\/p>\n

When done, report:
\n– Final counts per PARA bucket
\n– Any files you left untouched, and why
\n– Any new tags added to the vocabulary this session<\/p>\n

Guiding principle: I’m offloading the filing burden so I can focus on thinking. Use judgment. Ask when uncertain. Never force a note into the wrong bucket or invent tags just to clear the inbox.<\/p>\n

Claude then intelligently sorts, links, and moves files into the appropriate PARA folders based on that context.<\/p>\n


\n \"An<\/p>\n

<\/a><\/p>\n

Related<\/p>\n

\t\t
\n\t\t\t5 Useful AI Prompting Techniques You Should Know
\n\t\t<\/a><\/p>\n

Level up your AI game with these pro prompting strategies.\n<\/p>\n

Understanding the prompt<\/p>\n

Beyond basic organization, the prompt also creates two key files: an index file and a tags file.<\/p>\n

The index file acts as a map of the entire vault\u2014what exists, where it lives, and how it\u2019s structured. Instead of scanning every file, Claude can read this file to quickly understand the system, making the process faster and more token efficient. The tags file defines the set of tags used across the vault. This prevents tag sprawl by ensuring Claude reuses existing tags instead of creating new ones arbitrarily. If a new tag is needed, the prompt instructs Claude to ask for confirmation before adding it.<\/p>\n

Using Claude to \u2018distill\u2019 ideas<\/p>\n

Where messy captures become something worth keeping<\/p>\n

\"Using<\/p>\n

In Tiago Forte\u2019s framework, distillation is about progressive summarization\u2014revisiting your notes and highlighting key ideas layer by layer so future-you can quickly grasp the essence without rereading everything. I take a slightly different approach, though. For me, distillation is less about highlighting and more about processing notes\u2014individually or in groups\u2014and turning them into something more personal: a note that reflects my own thinking.<\/p>\n

To help me with this process, I use Claude as a sounding board. With access to my Obsidian vault, it can analyze what I\u2019m capturing and identify patterns in my interests. From there, it asks targeted questions that push me deeper into a topic or surface gaps and counterarguments I hadn\u2019t considered.<\/p>\n

This often turns into an engaging back-and-forth that helps me refine my thoughts. Claude then processes my new insights, condenses them into clear, structured notes, and saves them back into the vault for future reference.<\/p>\n

For this workflow, I\u2019ve had strong results with the following prompt:<\/p>\n

You are my thinking partner for distilling ideas from my Obsidian vault.<\/p>\n

Your job: help me review clusters of notes, spot patterns and gaps in my thinking, and extract atomic notes \u2014 sharp, single-idea crystallizations from messier source material. This is a conversational session. We’ll go back and forth until I decide to stop.<\/p>\n

PHASE 1 \u2014 Get oriented<\/p>\n

Ask me for the absolute path to my Obsidian vault. Once I share it, only read:
\n– \/_claude-index.md \u2014 my map of the vault
\n– \/_claude-tags.md \u2014 my approved tag vocabulary<\/p>\n

Keep both in working memory for the whole session.
\nDo NOT read all the notes cause that will eat into your entire context window.<\/p>\n

PHASE 2 \u2014 Offer two ways in<\/p>\n

Propose two modes and ask which I want to use today:<\/p>\n

MODE A \u2014 TOPIC<\/p>\n

Ask: “What do you want to talk about today? Anything on your mind?”<\/p>\n

If I’m unsure, suggest 1-2 concrete themes based on clusters you notice in the index. For example: “There’s a rich cluster forming around [X] \u2014 want to dig into that? Feels like there’s something worth unravelling there.”<\/p>\n

MODE B \u2014 SERENDIPITY<\/p>\n

Say: “I noticed some patterns worth surfacing.” This is NOT random selection. Scan the index for genuine signals:<\/p>\n

– Topics I’ve circled from many angles without landing a clear position
\n– Contradictions across my own notes
\n– Clusters where I have lots of raw material but no synthesis
\n– Themes I haven’t revisited in a while that may be relevant again given my recent notes<\/p>\n

Propose 2-3 of these with a one-line reason each, and let me pick one.<\/p>\n

Either way, land on a single focus theme for this session before moving on. If I pick multiple themes warn me about how it can dilute the distillation. But if I insist move ahead with multiple themes.<\/p>\n

PHASE 3 \u2014 Read the cluster deeply<\/p>\n

Using the index, identify every note related to the focus theme. Read them. Then present three outputs in one message:<\/p>\n

1. PATTERNS \u2014 recurring threads, how my thinking has evolved, where I’ve been consistent, where my stance has shifted over time.<\/p>\n

2. GAPS \u2014 questions I’ve raised but not answered, logical holes,unexplored angles, contradictions worth resolving.<\/p>\n

3. PROPOSED ATOMIC NOTES \u2014 single-idea crystallizations you’d extract from the messier source material. For each, show:
\n – Draft filename (kebab-case, descriptive)
\n – One-sentence core idea
\n – Which source note(s) it came from
\n – Proposed PARA location (usually the same folder as the dominant source; use judgment if the insight transcends one project)
\n – 2-3 tags from the approved vocabulary<\/p>\n

Do not save anything yet.<\/p>\n

PHASE 4 \u2014 Hold the conversation<\/p>\n

Now we talk. I’ll react to what you presented. Respond based on what I say:<\/p>\n

– If I confirm an atomic note (with or without edits): save it to the proposed PARA folder. Inside the note, add forward links to the source note(s) as [[source-name]] and to any related atomic notes you find in the index. Obsidian’s automatic backlinks will handle reverse linking,so don’t touch the source notes. Apply the approved tags. Append an entry to _claude-index.md.<\/p>\n

– If I push back or want to think more about an idea: don’t save. Keep discussing. Ask follow-up questions that push the thinking forward: “You keep circling X \u2014 is there something unresolved there?” or “These two notes contradict each other \u2014 which do you actually believe now?”<\/p>\n

– If I bring up a new angle mid-conversation: pull relevant notes from the index on the fly and work them in.<\/p>\n

– If a new atomic note needs a tag that isn’t in the approved vocabulary:propose the addition, wait for my explicit yes, then update_claude-tags.md before applying it. Do NOT add random tags without my permission.<\/p>\n

Keep the conversation going until I say we’re done. Don’t wrap up preemptively. If I go quiet, ask what else is on my mind.<\/p>\n

Guiding principles:<\/p>\n

– You are a sounding board, not an answer machine. Ask questions that help me think better. Point out what I might have missed.
\n– Atomic notes are permanent crystallizations. They should feel sharp,single-idea, and useful to my future self. Don’t save vague summaries or anything I haven’t explicitly confirmed.
\n– When in doubt, ask rather than assume. My thinking is the raw material; your job is to help me refine it, not replace it.<\/p>\n

Using Claude to \u2018express\u2019 creative ideas<\/p>\n

Where the whole system was pointing to<\/p>\n

\"Using<\/p>\n

The final stage is creating something\u2014an article, a project, a business strategy, whatever you\u2019ve been building toward. This is usually where things break down. Even if you\u2019ve been capturing and organizing consistently, finding the right ideas across months or years of notes is still a manual process.<\/p>\n

With this setup, you just tell Claude: I need to work on a project about [topic]. Look through my vault and help me brainstorm a fresh approach. Claude scans your notes, pulls out relevant ideas, and helps you connect them into something usable. Because it\u2019s grounded in your own material, the output reflects your thinking\u2014less generic, and more aligned with the ideas you\u2019ve actually developed over time.<\/p>\n


\n \"The<\/p>\n

<\/a><\/p>\n

Related<\/p>\n

\t\t
\n\t\t\tThis free plugin instantly turned my Obsidian notes into a beautiful website
\n\t\t<\/a><\/p>\n

You can convert your Obsidian vault into a website with one click.<\/p>\n

Claude and Obsidian finally make a second brain accessible to everyone<\/p>\n

I\u2019ve tried a lot of PKM setups over the years. Most of them collapsed under their own weight. This one hasn\u2019t\u2014because for the first time, I\u2019m not the one maintaining it. Claude and Obsidian handle the heavy lifting, turning what used to be a fragile system into something that actually holds up over time.<\/p>\n","protected":false},"excerpt":{"rendered":"Ever tried building a second brain, only to abandon it three weeks in? I have\u2014more times than I\u2019d…\n","protected":false},"author":2,"featured_media":29523,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[53,3154,182],"class_list":{"0":"post-29522","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-anthropic","8":"tag-anthropic","9":"tag-anthropic-claude","10":"tag-claude"},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/29522","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=29522"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/29522\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/29523"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=29522"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=29522"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=29522"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}