Obsidian has just released a highly ambitious update introducing Obsidian CLI its official command-
line tool which is not intended for everyday users
to interact with Obsidian through command-line inputs
but to fully embrace AI agents—this means
you can directly use Cloud Code Open Cloud
Lobster or even n8n workflows to execute
command-line commands and operate your Obsidia
This approach saves a significant number of tokens compared to the previous method
of letting AI directly read the notebook database
so you no longer need to worry about your agent consuming tokens
too quickly Today I'll guide you through learning it step by step
How to fully integrate your note library into AI agents using Obsidian CLI
Truly embracing 2026 the dawn of the agent era
Today's video is divided into the following sections
First I'll quickly show you how to enable Obsidian 's
CLI feature then use a few simple commands
to understand its principles and basic concepts
—don't panic at the sight of the command line
Because Obsidian CLI isn't meant for people—
it's for agents and AI
for workflows and code
What we need to do is give the agent tools the ability to use Obsidian CLI
After that I'll show you how to enable agents to use Obsidian CLI
We'll import the skill written by Obsidian's
CEO Steph into our agent enabling the agent to
directly access Obsidian via CLI and then I'll show you some practical application examples
How to make an agent operate our Obsidian via Obsidian CLI
After this I'll also show you another way to use Obsidian CLI
That is accessing Obsidian through code and n8n workflows
Additionally I'll introduce the core advantages of Obsidian CLI
And its significant importance in this era of intelligent agents
Finally I'll share some important
notes then introduce Obsidian 's
CEO Steph and a few other recent skills she's released
Steph's perspective on AI agents and her vision for Obsidian's future
key concepts covered in today's video
tool usage methods
code examples and prompt templates
I've organized everything into study notes
The video will share them with you at the end
So let's get started
First update Obsidian to the latest version
Click Check for Updates in the About section of the settings interface
Make sure your version
is 1.12 or higher
then scroll all the way down—there 's
a command-line interface toggle here turn it on and a window will pop up
Tell me to register Obsidian with PaaS
Then we click Confirm
So the Obsidian CLI feature is now enabled
Now let's open the command-line tool on our computer
Windows is CMD or PowerShell
Mac is Terminal
We run a basic command
Obsidian Daily
which creates today's journal
and after running this command we can see tha
My Obsidian interface automatically created today's journal
and opened it in the interface
and now we can clearly see a difference
this journal already contains content
so why is that
Since I have a diary template I can set my own
diary template by selecting Diary in the core
plugin section of the settings interface and
my template is located in the Templates folder
Let's take a look
in my journal template
there's a section for raw data
with three habit check-off tasks
namely exercise reading and meditation
this template is the one I previously explained to everyone
The heatmap tool for Obsidian was temporarily created
for this purpose thus it's
now clearly evident that using
the command-line tool operateObsidian directly triggered
Obsidian's internal mechanism—the template system
We know that previously using
tools like Cloud Code we could have
it create a note file but it was just creating a plain Mac file
In reality it accesses our operating system 's
file system equivalent to manually right-
clicking and creating a new file ourselves
while Obsidian CRI's underlying mechanism is IPC
That is progress and communication— it does not
directly operate disk files but instead sends commands
to the running Obsidian desktop application— meaning
any modifications made through code or an age
Changes will immediately take
effect in your UI interface in
real time and automatically maintain consistency within the
content I've shown you a structural diagram: previously
AI and Obsidian directly modified the files themselves but now
AI agents send commands to Obsidian's program process via
Obsidian CRI and then Obsidian
performs the operations on the notes
So one of the most important注意事项 with Obsidian CRI is
when using command-line commands
always ensure the Obsidian application is running
Let's run another command
Obsidian targets sort equals count
format equals json
This command lists all your notes' targets
and sorts them by frequency
You can see in my note library
the most frequent target is journal entries
Reading and efficiency—so what's the point
of such commands Here's an example to
make it clear: if you want to count the
tags in your notes or the original data
Previously your AI had to scan your entire note
library but now you can simply run a single command
to retrieve everything This makes it easy to
restructure your note library using AI or manage双向 links
Redesigning the tags and original data makes it
much more convenient and
AI doesn't need to read the
entire notebook library which greatly saves tokens
Now we'll immediately integrate AI-powered tools
Enable agents to use Obsidian CRI
We head to GitHub to search for the keyword Kapano
This person is Obsidian's CEO Steph
All of his social media accounts are under the name Kapano
Go to his GitHub homepage find the
Obsidian Skills repository
enter it download
all the skills inside the Skills folder
—this was covered in our previous video
How to use Agent Skills in Obsidian
If you haven't already
you can revisit my previous video
We'll install these Skills into our agent's toolkit
That way
Our agent has learned how to use Obsidian
CRI now let's immediately examine a
practical application example—here I'm using Obsidian CRI as the agent's tool
Come see how to showcase it You can also use tools like OpenCode or CloudCode
Of course you can use OpenCloud as the central hub
Or leverage Obsidian plugins as well
Integrating the agent's tools into Obsidian
You can review my previous videos for details
The reason for using Obsidian CRI
is still because it offers a generous amount of free token allowance
Recently I've been using OpenCloud's Lobster
and it's still very sensitive token usage I've
been raising lobsters lately and
it's left me completely drained spending a fortune in tokens
So here I'm sending a prompt to Gemini CRI
to retrieve the latest ten papers under
the AI directory on the Archive read
their information and then translate them
Organize everything into a table
then create a diary entry in my Obsidian
add the table to the diary
title it Today's AI Papers
then the agent begins working
Since we previously equipped the agent with the skill it will automatically invoke the Obsidian CRI skill to handle
this task the agent 's
execution requires a certain amount of time
We'll start by speeding up the video — you can
see that my Obsidian library does have a journal
template and currently the most recent journal
entry in the journal folder is from March 3rd
Today is March 8th
The journal template includes a habit check-in
Today's reflection and tomorrow's plan
This template was created when I previously explained the
Obsidian habit check-in heatmap to everyone
When the agent uses Obsidian CRI to create
a journal entry it triggers Obsidian 's
template system and directly creates the journal using this journal template
At the same time I set up a diary folder—my Diary folder—so
that日记 will automatically be
created inside the Diary folder
Obsidian CRI's alternative behavior is exactly the
same as clicking the New Journal button in Obsidian
Previously when we used AI agent tools to create
Obsidian notes we merely generated Markdown files
with no integration whatsoever with Obsidian's system
features now the journal has already been created
Let's take a look inside — it's clear the journal was created using a template and automatically
generated into the Diary folder — the table
lists the ten most recent AI papers from Archive
Also the table includes links to the papers clicking the link opens the Archive website
to view the papers That concludes the example
of using Obsidian CRI in the AI agent tool
Next I'll quickly show you how
to use Obsidian CRI in code
or n8n's command line node — just take a quick look at this part
If you'd like to try webcoding remember to include
the official Obsidian CRI documentation with
your request to the AI since the AI's training data definitely doesn't contain this material
You don't need to panic at the mention of code—
just a simple prompt is all it takes to get AI to
generate code like this and I personally use AI
to create plenty of such code on a regular basis
This is the kind of web coding that ordinary people
should really learn because it genuinely helps boost
our efficiency in everyday tasks like file handling
data analysis and working with agents and workflows
On the contrary I think it's really
unreliable to expect someone with zero
programming experience to use AI to build
an entire project or game from scratch
I've also shared the prompts used to generate this code in my notes—feel free to
use them directly—just remember to install
Python on your local machine beforehand
Let's take a look at the Python code this AI has
written for us — it similarly calls the archive'
s link to retrieve information on the latest 10
AI papers then organizes it into a markdown table
Then create a note using Obsidian CRI
then add the table to the note This
time when creating the note use this
weekly plan and sub-board template
In Obsidian CRI's commands you can
create a note while specifying a
template and we'll name the note "2026
March Week 2 AI Learning Materials"
Then we directly run this Python code
and we can see the result
I'll briefly show you the part of the Python code
that calls the Obsidian CRI command
It's actually quite simple
You just need to send the Obsidian CRI documentation to the AI and it will know
how to call the command-line to execute
Obsidian CRI commands within Python code
We can see the note has been successfully created
automatically applying our weekly plan and sub-
board template with the AI-generated paper content
also added to the note—similar information capture
Data retrieval and daily workflow tasks
can all be integrated into Obsidian CRI to
seamlessly connect with your Obsidian workflow
similarly if you've deployed n8n locally
To make n8n work with your
Obsidian CRI notebook library it's
very simple: add an execute
command node to the workflow panel
Then simply run the Python script you want
For those unfamiliar with n8n
you can revisit my previous videos on integrating n8n with Obsidian and Notion
Obsidian CRI's greatest advantage is reducing the token consumption
of AI agents Previously AI had to scan your entire note library
Now you can directly query the internal structure and indexes of your note library via CRI
— executing a single command to find orphaned notes in your library requires only 100 tokens
Previously scanning an entire note library might have required millions of tokens
Additionally Obsidian CRI enables AI to better perceive the structure of your note library
Including dual links tags plugin status and
more — this was previously very difficult
to achieve — and by using CRI to move
files data consistency can be maintained
Internal links are automatically updated when
moving files preventing broken links Modifying YAMO's
original data requires only a single command AI does not need to read the entire file
That means the release of Obsidian CRI signifies Obsidian's full embrace
of the agent era and solves the biggest
issue of excessive token consumption
Obsidian's CEO Steph previously stated in an interview
that they wouldn't launch an official AI tool
like Notion because Obsidian follows a local-first
principle keeping all data firmly in users' hands
Privacy-focused with users in control of how AI is used
and when the Agent Skill standard was just released in early 2026
Steph immediately published three Obsidian Skills
I've mentioned Steph before—he's the creator of the well-known
Obsidian theme skin Minimo and later
unexpectedly became Obsidian's CEO
His GitHub page already has five Obsidian Skills published—three
that I introduced in my previous
video plus this Obsidian CRI skill
Another is the web content extraction skill
which allows you to pull content from web
pages and convert it into clean Markdown
format—perfect for browsing information online
You can directly pull the content into
our Obsidian notebook library now everyone
can install these few skills into your
agent and enable the CRI feature in Obsidian
Let the agent take over your knowledge base
My advice to everyone is
don't fixate solely on the agent
because as I just mentioned
CRI can be used in both code and n8n workflows—this
is mainly because I've recently been using OpenCloud 's
Lobster and during lobster farming I've realized
it actually consumes a significant number of tokens
I'm sure students who like
me have been raising lobsters
using OpenCloud have had firsthand experience with
excessive token consumption
so I still recommend that for
more deterministic tasks
workflows and code should be used
Instead of using an agent
otherwise your token
usage will skyrocket Also if your primary
use case is Obsidian
I recommend trying Gemini CRI and OpenCode—these two agent tools
They still have a generous free token allowance
That's all for today's video
If you encounter any issues
while using Obsidian CRI feel
free to leave me a message
Today's video's notes prompts and code are all
available for download on
my personal homepage—you can
find my homepage in the
info section of my
channel Don't forget to like and subscribe Thanks everyone