Use AutoGen Studio and add a new skill to store notes in Obsidian

autogen studio is a UI maintained by the autogen team we can run it locally on our device in this tutorial we will add a new skill using autogen Studio to store some noes in obsidian we will place the python function needed to store the markdown file and the note content in a new skill next we will use the new skill in a new agent and a new agent in a new workflow in the playground we select the workflow and can ask autogen to use the new skill this is similar to open AI function calling and enable us to use the workflow and storing information as marked down in our note taking app when we switch to obsidian we see our not taking agent used the new skill and stored some information in markdown format in the app this help us to use autogen to automate our research and not taking workflow we use in this tutorial obsidian there are many not taking apps out there from ever node to notion but obsidian has a huge advantage over them as it is local on our device furthermore it builds a craft to visualize the the relationship between our notes and thoughts so let's dive in to see how we can use autogen and obsidian together using a custom new skill in autogen studio we start with creating a new directory in our project folder next we switch to the new directory and from inside the directory we start Visual Studio code in Visual Studio code we create a new virtual environment and after the new virtual environment is created we activate it make sure the name of the virtual environment appears before the prompt next we assign our open AI key to open aior API key in Windows we use set to do this but if you are on the Mac you must use export instead of set now we are are ready to install autogen Studio we use pip install autogen Studio to install it it takes a while to set up all of the packages after the packages are installed the prompt comes back and we can clear the screen and start autogen Studio to start autogen Studio we type autogen Studio ui-- port 8881 when we navigate to the address we see the autogen studio UI in the build tab we can add skills these skills can be added to agents which in turn can be added in workflows after we defined our workflows we can test them in the playground in the playground we can create sessions and for each session choose a workflow from the drop-down list if you we are happy with the execution and result we can publish the workflow to the gallery for a quick start and testing the API key we can simply use one of the provided example prompts here we choose the markdown prompt which sends this out the top five rivers in Africa and their length and return it as a markdown table do not try to write any code just write the table autogen generates the answer and return it as a markdown table although we can publish this to our Gallery we want to add a skill to autogen to store it in our not taking application obsidian which acts as our second brain as mentioned before obsidian is one of the best and most popular not taking apps and with a very active Community behind it this community has written more than one ,300 plugins for obsidian from to-do and calendar plugins to Canan and mind maps you can learn more about obsidian using the obsidian help here you can explore the core plugins when you expand the interactive graph you can see the relationship between the nodes you can hover over a node to focus on the relationship of that particular node or click the node to see the underlying information you can search for tle cust or PKM for personal Knowledge Management to learn more about this subject back in Visual Studio code we can see the look of our last prompt and see the markdown table generated we can use the visual studio code to create the function to be used in the skill we use the IDE for syntax highlighting and an environment to test the script it doesn't have to be in our project but to keep everything in place I put it here we create a python file and paste the code for the skill in the file to run the script we have to choose an interpreter let's review The Script first we import the required packages then we Define a function for Simplicity we hardcoded the root to the obsidian Vault and assign it to obsidian deer as obsidian works with markdown files and the llm may not give a name with a MD extension we will add it if needed then we join the obsidian rout and the name of the file with the extension MD to get the full path and finally write the content to the file we can test the function and run the script with test as file name and a simple markdown string as content when we switch to obsidian we see the extension is added and the file is listed in obsidian the markdown is rendered and we see a bold and a highlighted part in the string we can switch to the source mode and see the content of the file back to visual studio code we remove the part to call the function and copy the rest to be used in our new skill in autogen studio we navigate to build and add a new skill we delete the sample skill and paste our custom function we can simply use the function name as the name of the skill and save it next we add an agent we name it no taking agent this name is important as we will use it in the next step in our workflow we give a description and add the new skill from the drop- down list and save the agent the next step is to define a workflow we name it not taking agent workflow and give it a description here is important to use our not taking agent as the receiver the sender is the user proxy we add the skill and save the workflow now it's time to to test the workflow so we navigate to The Playground here we still see our last session and the list of the African Rivers we create a new session and select the note taking agent workflow from the drop down list with the new workflow selected we enter the prompt list out five open AI models and a context window in a markdown table and stored it as a note in obsidian store each model as a note to going back to the terminal in Visual Studio code we can see the communication under the hood but as we have now the autogen studio we have a much better user interface we can see the steps and result in the much better way switching to obsidian you see the note containing the markdown table with the information and for each model is separate note with some information about the model we can rename doly and change the cell in the database it is very easy to create a link to an existing note back in doly we can create a link back to the table when we open the graph we see a relationship between the table and the doll e node we have started to build our second brain you can click on the notes in the graph to get to the note itself this is the beginning of our personal Knowledge Management or PKM we can serf the web and find more related information and extend our knowledge and our table with the new table tool in obsidian we can easily modify our markdown table move rows create new columns and move the columns and and continue with our research I highly recommend taking the time and build a clean organized second brain as for Learning and organizing all of the AI information one brain is simply not enough so to wrap it up working with autogen is now much easier using the autogen studio as the user interface we can easily extend it using our new custom SC dets if you need more information regarding autogen Studio navigate to autogen and check the autogen Studio blog post good luck using autogen Studio and creating new skills

Channel: business24_ai

Views: 5,776

Keywords: autogen, AutoGen Studio, obsidian

Id: OMa9_xuFG0s

Length: 10min 46sec (646 seconds)

Published: Sun Dec 31 2023