OpenDevin Tutorial (Open-Source Devin) - Build Entire Apps From a Single Prompt

Matthew Berman
30 Mar 202414:58

Summary

TLDRこの動画は、オープンソースのDevonツール「Open Devon」の紹介とデモンストレーションです。デモでは、シンプルな計算器アプリを作成し、ターミナル、ブラウザ、コードエディタを統合したインターフェースで動作デモンストレーションを行いました。Open DevonはGitHubでスター数多く、迅速に成長しています。また、Pythonパッケージ管理や環境設定に関する問題も解決方法を提案しています。

Takeaways

  • 🌟 オープンソース版のDevonがリリースされ、名前はOpen Devonです。
  • 📱 Open Devonは非常に印象的なインターフェースを持ち、ターミナル、ブラウザ、コードエディタ、エージェント対話が統合されています。
  • 🔧 Open DevonはオープンAIとCloud 3をサポートしていますが、オープンソースのローカルモデルも簡単にプラグインできます。
  • 💻 Open Devonを使ってPythonでシンプルな計算機アプリを作成することができました。
  • 🖥️ 計算機アプリはローカルホストで実行され、Webインターフェースも作成・テストができました。
  • 🔍 Open Devonはまだ開発の初期段階にありますが、非常に使いやすく、進化速度も非常に速いです。
  • 🚀 GitHubでトレンド入りし、8,500以上のスターを獲得しています。
  • 🛠 インストール手順はVisual Studio Codeから始まり、新しいconda環境の作成やDockerの使用が含まれます。
  • 🔑 OpenAIのAPIキーを設定する必要があり、.envファイルを作成して安全に管理することが推奨されています。
  • 🐞 Open Devonはまだいくつかのバグがあり、完全に安定していない可能性がありますが、コミュニティの協力によって迅速に改善が期待されています。
  • 🎉 ビデオの制作者は、Open Devonが開発者や非開発者にとって非常に生産的なツールになることを楽しみにしています。

Q & A

  • DevonとOpen Devonの違いは何ですか?

    -Devonは閉鎖源のコードベース構築ツールで、デモで展示されました。一方、Open DevonはDevonのオープンソースバージョンで、同じ機能を提供しています。

  • Open Devonのインターフェースにはどのような機能がありますか?

    -Open Devonのインターフェースには、ターミナル、ブラウザ、コードエディタ、およびエージェント対話が含まれています。

  • Open Devonをインストールするにはどうすればいいですか?

    -Open Devonをインストールするには、GitHubからリポジトリをクローンし、Visual Studio Codeでプロジェクトを開きます。その後、新しいconda環境を設定し、必要なパッケージをインストールしてDockerイメージを取得します。

  • Open Devonで使用されるモデルは何ですか?

    -Open Devonは現在、OpenAIとCloud 3のモデルをサポートしていますが、オープンソースのローカルモデルも簡単にプラグインできます。

  • Open Devonを使ってシンプルな計算機アプリを作成する方法はどのように進んでいますか?

    -Open Devonを使って計算機アプリを作成するためには、シンプルなプロンプトを与え、ターミナルから指示に従ってコードを生成し、ブラウザで結果を確認します。

  • Open DevonがGitHubのトレンドにあることはどのくらいの意味ですか?

    -Open DevonがGitHubのトレンドにあるということは、そのプロジェクトが非常に人気があり、開発が活発に進んでいることを意味します。

  • Open Devonの開発においてまだ解決されていない問題は何ですか?

    -Open Devonはまだ開発の初期段階にあり、いくつかのバグや機能がまだ動作していない可能性があります。しかし、進歩のペースは非常に速く、問題はすぐに解決される可能性が高いです。

  • Open Devonでローカルモデルを使用するにはどうすればいいですか?

    -Open Devonでローカルモデルを使用するには、適切なAPIキーをエクスポートし、環境変数または.envファイルに保存する必要があります。

  • Open Devonで遭遇したエラーを解決するための一般的なアドバイスは何ですか?

    -Open Devonで遭遇したエラーを解決するためには、AIと相談するか、GitHubのIssueページをチェックすることが一般的です。それにより、解決策を見つけることができます。

  • Open Devonを使って開発者や非開発者がどのように生産性を向上させることができますか?

    -Open Devonを使用することで、開発者はコードをより迅速に構築でき、非開発者もプログラミングの概念を理解しやすくなります。これにより、生産性が向上し、効率的な開発が可能です。

  • Open Devonの将来展望は何ですか?

    -Open Devonの将来展望は、より多くの機能の追加、バグの解決、そして開発コミュニティからの貢献によってますます充実し、生産性を向上させることが見込まれます。

Outlines

00:00

🚀オープンソース版Devonの紹介とインストール方法

この段落では、数週間前に見たインパクトのあるDevonのデモとそのオープンソース版であるOpen Devonについて紹介されています。デモでは、代理者がプロンプトを受け取り、完全なコードベースを作成することができ、ターミナル、ブラウザ、コードエディタ、エージェント対話が含まれているインタフェースが示されています。しかし、クローズドソースであったDevonの大きな欠点と、オープンソース版のリリース、およびそのインストール方法と使用方法について説明されています。

05:01

🛠️DockerとPython環境設定の詳細

この段落では、Dockerのインストールと使用方法、Python仮想環境の設定、およびOpen Devonプロジェクトの要件をインストールするための手順が詳細に説明されています。Dockerの利点、Dockerイメージのダウンロード方法、Open AI APIキーの取得と設定方法、ワークスペースディレクトリの設定、Pythonのバージョン確認とPipの使用、Rustのインストール、およびノード.jsとnpmのインストールについても触れられています。

10:01

🌐バックエンドとフロントエンドの実行とトラブルシューティング

この段落では、Open Devonのバックエンドとフロントエンドの実行方法、問題が発生した際のトラブルシューティング手順が説明されています。バックエンドの実行にuicornを使用し、フロントエンドの実行にはnpmを使用しています。また、実行中に遭遇した問題、例えばフリーズやエラー、およびそれらを解決する方法についても言及されています。最後に、任意のモデルを使用する方法、特にローカルモデルを使用する方法についても説明されています。

Mindmap

Keywords

💡Devon

Devonは、プロンプトを与えられたりコードベースを構築するインタラクティブなインターフェースを持つデモです。このデモはオープンソースのバージョンであるopen Devonとして導入され、ビデオの主題を形成しています。

💡open Devon

open Devonは、Devonのオープンソースバージョンで、ユーザーがコードを構築し、プロジェクトを進めることを助けるツールです。ビデオの中心的な役割を果たしています。

💡interface

インターフェースとは、ユーザーがソフトウェアやシステムとやり取りするためのメカニズムです。ビデオでは、Devonのインターフェースがターミナル、ブラウザ、コードエディタ、エージェントの対話を統合していることが強調されています。

💡calculator app

計算器アプリは、ビデオで示されたopen Devonの使い方の例です。このシンプルなPythonアプリケーションは、open Devonを使って開発され、その機能と使いやすさを示すために用いられました。

💡GitHub

GitHubは、ソフトウェア開発者がコードをホストし、バージョン管理、チームでの共同作業を行えるプラットフォームです。ビデオでは、open DevonのプロジェクトがGitHubでホストされていることが言及されており、ユーザーがこのプロジェクトにアクセスし、インストールする方法が説明されています。

💡Docker

Dockerは、アプリケーションをコンテナとしてパッケージ化し、異なる環境で一貫性を保ちつつ実行できるようにする技術です。ビデオでは、Dockerがopen Devonのインストールと実行において役立っていることが述べられています。

💡API key

APIキーは、ソフトウェアアプリケーションがAPI(Application Programming Interface)を利用するために必要な認証情報を提供するものです。ビデオでは、open Devonでopen AIを利用するためにAPIキーを設定する必要があることが触れられています。

💡Python

Pythonは、世界中で広く使われているオープンソースのプログラミング言語で、高い可読性とシンプルな構文が特徴です。ビデオでは、Pythonを使ってシンプルな計算器アプリを作成し、open Devonの機能を紹介しています。

💡Visual Studio Code

Visual Studio Code(VS Code)は、Microsoftが開発した無料のオープンソースのコードエディタで、多くのプログラマーが愛用しています。ビデオでは、VS Codeを使ってopen Devonをクローンし、設定する方法が説明されています。

💡Node.js

Node.jsは、サーバーサイドのJavaScriptランタイムであり、ウェブ開発において人気のあるプラットフォームです。ビデオでは、Node.jsを使用してopen Devonのフロントエンドをインストールし、実行する方法が示されています。

💡Uvicorn

Uvicornは、PythonのASGI(Asynchronous Server Gateway Interface)サーバーであり、非同期Webアプリケーションを効率的に実行できるように設計されています。ビデオでは、Uvicornを使用してopen Devonのバックエンドサーバーを起動する方法が紹介されています。

Highlights

The introduction of an open source version of Devon, called Open Devon, which allows users to build entire codebases from prompts.

Open Devon's impressive interface that includes a terminal, browser, code editor, and agent dialogue.

The capability of Open Devon to write a simple calculator app in Python upon a user's request.

The integration of different models in Open Devon, such as Open AI and Cloud 3, with the possibility to plug in an open source local model.

The demonstration of Open Devon's ability to create an HTML interface for the calculator app, showing its versatility.

The cost-effectiveness of using Open Devon for building apps, as it requires minimal tokens.

The rapid growth and popularity of the Open Devon project, with over 8,500 stars on GitHub and being the number one trending app.

The guide on how to install Open Devon, including cloning the repository and setting up the environment using Visual Studio Code.

The requirement of Docker for easier installation and running of code environments in a Dockerized environment.

The process of obtaining and using an Open AI API key for Open Devon.

The potential issues with Python package management and environment management that might be encountered during installation.

The solution to installing Rust and its dependency oJson, which was a challenge faced during the setup.

The method of using an .env file for storing environment variables, which is recommended over temporary exports.

The step-by-step guide on installing front-end packages with Node and npm, and running the server with uicorn.

The capability to use any model with Open Devon, including local models, by exporting the necessary API keys and settings.

The acknowledgment of some bugs and issues in Open Devon, but also the rapid progress being made by the developers.

The encouragement for users to contribute to the Open Devon project by creating issues on their GitHub repository.

Transcripts

play00:00

do you remember Devon that incredible

play00:01

demo that we saw just a couple weeks ago

play00:04

where you gave the agents a prompt and

play00:06

they built out entire code bases and I

play00:08

know we've seen stuff like that before

play00:10

but what really set Devon apart was the

play00:13

fact that it had this incredible

play00:14

interface that included the terminal the

play00:16

browser the code editor and all the

play00:18

agent dialogue and it was really

play00:20

impressive but it had one major flaw it

play00:22

was completely closed source and not

play00:25

more than a couple days after that demo

play00:27

went viral now we have a completely open

play00:30

source version of Devon called open

play00:32

Devon and today I'm going to show it to

play00:35

you and that's what you're looking at

play00:36

right now so I'm going to show it to you

play00:38

I'm going to show you how to use it and

play00:39

I'm going to show you how to install it

play00:41

so this is it and I already set it all

play00:43

up so let me show it to you first so it

play00:45

says hello I'm Devon and I asked it

play00:48

write a simple calculator app with

play00:50

python so it started writing a new task

play00:52

here on the right side you can see the

play00:54

terminal very akin to the original Devon

play00:57

we have the planner over here we have a

play00:59

code editor where you can actually see

play01:01

all of the code and then we have the

play01:03

browser over here you can select the

play01:05

different models that you want to run so

play01:07

right now it supports open Ai and Cloud

play01:09

3 but you can easily plug in an open

play01:11

source local model as well and then over

play01:13

here we have the browser and you can

play01:15

choose between Lang chains agent and

play01:17

code act agent and to be honest I don't

play01:20

actually know the difference between

play01:21

these two I haven't had a chance to test

play01:23

code act agent so I'm using langing

play01:25

chains agent but let's continue so

play01:28

starting a new task then we go over to

play01:29

the the terminal and we can actually see

play01:31

so command LS there it is it looks where

play01:33

it's at then it's reading from the

play01:36

app.py it seems there's already a file

play01:39

that does this it's a calculator app and

play01:41

so on and so forth and you can see all

play01:42

the output here it even tested it for me

play01:46

which is really nice and then after all

play01:48

that which really just took a few back

play01:50

and forths and it's a very simple app

play01:52

build a calculator app all done what's

play01:54

next on the agenda I iterated on it so

play01:56

now make an HTML interface for the

play01:59

calculator so starting a new task and

play02:01

then it went back and forth spun up

play02:03

Local Host even tested it made sure it

play02:05

all worked and then it was done so all

play02:07

of this was actually pretty inexpensive

play02:10

tokens wise um but it was really

play02:12

impressive and you can run the

play02:13

calculator like this so python

play02:15

calculator. piy or you can spin up a

play02:18

server and here's the calculator this is

play02:20

what it built for me with just that

play02:22

simple prompt so you put in your number

play02:24

right here put in another number I'll

play02:26

say four calculate very very basic but

play02:30

the point is it works now I'm back at

play02:32

open Devon and I am running it on Local

play02:35

Host so this is running locally I am

play02:36

using gp4 although I could easily swap

play02:40

out an open source model which I'll show

play02:42

you in a bit now the important thing to

play02:43

remember is this project has not been

play02:45

around long I'm talking days so there

play02:48

are still some bugs some features still

play02:51

don't work but it is very usable and the

play02:55

rate of progression and new features

play02:58

being added is super impressive

play03:00

impressive so this is the project open

play03:02

Devon it has over 8 and half thousand

play03:04

stars already and if you look at GitHub

play03:07

trending it is the number one trending

play03:09

app on GitHub so this is going to grow

play03:12

quickly and if I scroll down a little

play03:14

bit there's another open source version

play03:16

of Devon project called DEA although I

play03:18

have tried every single day to get this

play03:20

working and I can't so I'm going to keep

play03:22

trying as soon as I can I'll make a

play03:24

tutorial video for that but I am able to

play03:26

get open Devon working and it works

play03:28

really well so enough talk let me show

play03:31

you how to install it I ran into a bunch

play03:33

of issues hopefully I will show you how

play03:36

to solve all of them and most of the

play03:38

issues actually have nothing to do with

play03:40

Devon they have to do with python

play03:42

package management and environment

play03:43

management which you know is the bane of

play03:46

my existence so open a visual studio

play03:49

code and that's where we're going to

play03:50

start click on this button to toggle the

play03:52

panel and we're going to open up our

play03:53

terminal and what we're going to do is I

play03:55

like to put stuff on my desktop when I'm

play03:57

first playing around with it so we're

play03:58

going to CD to the desktop now switch

play04:00

back to the open Devon GitHub repository

play04:03

you're going to click this green code

play04:05

button right there and we're going to

play04:06

copy the GitHub repo URL now we're going

play04:08

to switch back to our terminal and we're

play04:10

going to type get clone and then paste

play04:12

in that URL and then hit enter and

play04:15

that's it we've cloned it to our desktop

play04:17

next we're going to CD into open Devon

play04:20

and next we're going to click this

play04:22

little button right here Explorer we're

play04:24

going to open folder select the desktop

play04:26

and then we're going to select open

play04:27

Devon and now we have the open Deon

play04:29

project open in Visual Studio code all

play04:32

right now that we have that going let's

play04:34

spin up a new cond environment so we're

play04:36

going to do condac create DN o for open

play04:39

Deon python equals 3.10 and we're going

play04:42

to hit enter now I already have an

play04:43

environment named that because I've gone

play04:45

through this once to make sure it all

play04:46

works before I show it to you so you're

play04:48

not going to see this but go ahead and

play04:50

install it just hit enter all right once

play04:52

that's done we're going to grab this

play04:54

Command right here cond to activate OD

play04:56

copy paste it and it should say OD right

play04:58

here it may not in your terminal if your

play05:01

terminal structure is a little different

play05:02

but for me I show it right there so

play05:04

there we go we have o activated the next

play05:07

thing you're going to need is Docker and

play05:09

I'm really glad that they use Docker

play05:11

because it makes the entire installation

play05:13

much easier and you can actually run

play05:14

these code environments in a completely

play05:17

dockerized environment so to check if

play05:19

you have Docker you're going to type

play05:20

Docker PS and I do and there it is

play05:23

however when you run Docker PS you might

play05:25

get Docker is not recognized and if

play05:27

that's the case you need to download

play05:28

docker so you're going to come to docs.

play05:31

do.com sengine install and you're going

play05:34

to look for the docker desktop client

play05:36

that matches your operating system so

play05:39

I'm on a Mac so I click right there once

play05:41

you do that it'll download and

play05:43

everything else is really just drag and

play05:45

drop or kind of clicking through an

play05:46

interface it's very very easy you don't

play05:48

need to do anything in console once

play05:50

you're done with that open up vs code

play05:52

again and you're going to type Docker PS

play05:55

and now you should see at least this top

play05:57

row right here container ID etc etc so

play06:00

the next thing we're going to do is pull

play06:02

the docker image and again this makes

play06:04

everything really easy so we're going to

play06:05

type Docker pull

play06:08

gc.

play06:10

iops sandbox colon

play06:13

v0.1 and hit enter and there we go it's

play06:16

150 megabytes downloads quite quickly

play06:19

extracts and we're done so that worked

play06:21

perfectly okay next we need to export

play06:24

our open AI API key so to start we're

play06:27

going to use open AI but I'll show you

play06:29

how to set up a local model towards the

play06:30

end of the video so if you don't already

play06:33

have an open AI account go ahead and

play06:35

sign up you need a developer account

play06:36

platform.

play06:38

open.com ai- Keys you're going to click

play06:41

create new secret key and I'm going to

play06:43

type odop Devore YT so I know it's for

play06:48

YouTube and I will revoke this key

play06:50

before publishing this video click copy

play06:52

go back and we're going to export it

play06:54

just like that and then hit enter okay

play06:57

now we've exported it and basically what

play06:59

export it does is it saves it as an

play07:01

environment variable that we can use

play07:02

with this software however the better

play07:04

way to do it is to actually create a EnV

play07:07

file and store it there but I'll leave

play07:09

that for you to do the next thing we

play07:11

need to do is set our workspace

play07:12

directory and so what I'm going to set

play07:14

it as is export workspace Ford equals

play07:18

squiggly line/ desktop slop Deon so I'm

play07:22

going to keep the workspace in the open

play07:24

Devon folder just to keep it all in one

play07:26

place so go ahead and hit enter there

play07:28

all right now we're going to going to

play07:29

install the requirements and this is

play07:31

where I started to run into some

play07:32

problems so I may not run into it again

play07:34

just because I've solved them already

play07:37

but if I do I'll show it to you and even

play07:38

if I don't I'll show you the problems I

play07:40

had and I'll show you how I solved them

play07:42

so we're going to type which python okay

play07:44

and this is only because we want to make

play07:47

sure that we're using the correct python

play07:49

for installing with Pip so we grab the

play07:52

python version we're using then we

play07:54

simply paste that in type-m PIP install

play07:58

dasr requirements. txt then hit enter so

play08:01

one of the issues that I faced is that

play08:03

this project requires rust and

play08:05

specifically the dependency o Json

play08:09

requires rust and I didn't have it

play08:11

installed and so I had an error here so

play08:13

I didn't have it this time so this is

play08:15

going to be a little bit of behind the

play08:16

scenes but anytime that I do a tutorial

play08:18

video I go through it once without the

play08:20

camera recording and I document every

play08:22

step along the way and I also document

play08:25

any errors or bugs that I run into so

play08:27

that when I go to record I it doesn't

play08:29

take me forever so I did have to install

play08:32

rust and to do that I used this command

play08:36

curl d-pro parentheses equals htps D-

play08:42

tlsv1.2

play08:44

dssf and then so on and by the way I'll

play08:47

drop all these commands in a GitHub gist

play08:49

just so you have them and you don't have

play08:50

to try to copy them and the next thing I

play08:53

had to do was restart the terminal so

play08:54

keep that in mind so one other thing I

play08:56

want to point out another issue that I

play08:58

ran into is the o Json issue and to

play09:01

First fix it I installed rust and then I

play09:03

ran into another issue with o Json and

play09:06

to fix that I did this pip uninstall or

play09:09

Json and then I installed it again using

play09:11

this longer command which basically

play09:14

installs the binary version that is

play09:16

specific to my Apple silicon and that

play09:19

was the problem and all of these

play09:21

problems might be very specific to my

play09:22

machine and you might run into other

play09:24

problems I recommend Consulting Ai and

play09:27

it will help you just copy paste

play09:28

whatever issue you're running into and

play09:30

it usually will give you some pretty

play09:31

good suggestions so that command is PIP

play09:34

install D- noach d-- only binary colon

play09:39

all colon o Json and once I did that it

play09:42

finally worked and if you do restart the

play09:45

terminal you need to export the open AI

play09:47

API key again because as soon as you

play09:49

restart the terminal all of those

play09:51

temporary environment variables are

play09:53

wiped that's why using the EMV file is

play09:56

always better all right now that that's

play09:58

all done we're going to try try to spin

play09:59

up the server and hopefully it works it

play10:01

uses uicorn and this is the back end so

play10:04

we need the back end and the front end

play10:05

working so let me show you what to do

play10:07

here uicorn open de. server. listen

play10:11

Colona d-port 3000 now let's see if I

play10:14

run into an issue last time when I tried

play10:16

to spin up the server it would just

play10:18

completely freeze and so I actually had

play10:20

to restart the terminal anyway so we'll

play10:21

see if we have to do that here all right

play10:23

so it is looking like it's hanging again

play10:26

unfortunately so what we're going to do

play10:28

is hit controll C to try to quit out of

play10:30

here although I think it's completely

play10:32

Frozen so we're going to have to hit

play10:33

contrl Z and that'll force quit it and

play10:36

so let's try it again and if this

play10:39

doesn't work I'm going to try restarting

play10:40

the terminal completely again all right

play10:43

maybe I'm being a bit impatient but I

play10:45

don't think it's working so I'm going to

play10:46

hit oh I spoke too soon look at that so

play10:49

the second time it did work maybe it's

play10:51

doing some downloads in the background

play10:52

I'm not sure but it did work on the

play10:54

second go so we have uicorn running at

play10:57

Local Host 3000 perfect

play10:59

now what we're going to do is we need to

play11:01

now install and spin up the front end so

play11:04

we click the little plus button right

play11:06

here we wait till we get our new

play11:08

terminal up and running we're still in

play11:10

the open Devon folder and we still have

play11:12

OD cond environment running so just

play11:14

verify those things now we're going to

play11:17

CD into the folder called front end and

play11:21

we're going to be using node to install

play11:23

it an npm and if you don't have node if

play11:25

you don't have npm you need to go Google

play11:27

that and or use Claud or GP or something

play11:30

and just get those two things installed

play11:32

it should be pretty straightforward I

play11:34

believe if you're using a Mac you can

play11:35

even use Brew so you could do like Brew

play11:37

install npm and it should work I believe

play11:41

all right so there it is so that would

play11:43

work so now we have npm installed all

play11:45

right so now that we have node installed

play11:48

let's do npm install and that's going to

play11:51

install all the front-end packages now

play11:53

luckily I have much fewer issues using

play11:57

npm and the whole node e ecosystem and

play12:00

package management with node much fewer

play12:02

issues than I do with python so

play12:04

hopefully you don't run into anything

play12:05

okay now that we have all of the node

play12:08

packages installed we are simply going

play12:10

to spin up the node server now so npm

play12:13

run start-- space- dport space 3001 and

play12:17

then hit enter and that's it we should

play12:19

be up and running now let's give it a

play12:21

try so I'm going to click on this Local

play12:23

Host right there actually I'm going to

play12:25

hold down command then click on the

play12:26

local host and there we go open Dev

play12:29

it worked wonderful so it takes a few

play12:32

seconds to initialize the agent and I'm

play12:34

going to switch to GPT 4 over here and

play12:36

there we go hello I'm open Devon what

play12:38

would you like me to build so I'll say a

play12:40

simple website that says hello world and

play12:42

now we'll see it working a little bit

play12:44

starting new task we can also click over

play12:46

to the planner now I've noticed the

play12:48

planner doesn't really update that often

play12:50

or maybe even at all um maybe that's a

play12:53

little buggy I've also found the browser

play12:55

doesn't really work all that well to be

play12:57

honest the terminal seems to work great

play12:59

and the code editor definitely works so

play13:02

I mean there is the code there's the

play13:03

hello world HTML file perfectly done so

play13:06

here we go it's starting up a server all

play13:08

by itself and it visited Local Host 8000

play13:13

so if I actually go over to the browser

play13:14

it did switch and go over to Local Host

play13:16

8000 so it kind of works but there's

play13:19

some little bug and it doesn't work

play13:21

completely yeah and if I go back to the

play13:23

logs from the back end I can see that

play13:26

there was an error here and it exited so

play13:28

that's it it so definitely still buggy

play13:31

but they're making a ton of great

play13:32

progress now let me show you how to use

play13:35

basically any model including a locally

play13:37

run open source model so if you wanted

play13:39

to use Claud you just export these

play13:42

things right here so the llm API key and

play13:44

then the llm model you export this and

play13:47

you do that from terminal now if you did

play13:50

want to use a local model you do llm

play13:52

base URL and you change it to Local Host

play13:55

3000 and then you can use LM Studio you

play13:58

can use ol llama you can use anything

play14:00

you want as long as it exposes an open

play14:02

AI compatible API endpoint and you can

play14:06

even select llama 2 for your embedding

play14:08

model which is really cool so you could

play14:11

technically get this to be completely

play14:13

local if you wanted so they are truly

play14:16

trying to mimic what Devon has done and

play14:19

Devon is super impressive it is

play14:21

definitely not the first time we've had

play14:23

coding assistance it's actually far from

play14:25

it but it is one of the most if not the

play14:27

most polished user interface that I've

play14:29

seen so I'm really excited for open

play14:32

Devon I've tried this other project DEA

play14:34

a bunch and I haven't gotten it to work

play14:36

but open Devon works pretty darn well so

play14:38

give it a try create issues on their

play14:41

GitHub repository as you come across

play14:42

them contribute if you're open to that

play14:45

and open Devon can be something really

play14:47

special that helps developers and even

play14:49

non-developers be really productive at

play14:52

building code if you liked this video

play14:54

please consider giving a like And

play14:55

subscribe and I'll see you in the next

play14:57

one

Rate This

5.0 / 5 (0 votes)

Related Tags
オープンソースDevonプログラミング開発支援ツール紹介インストールガイドコードエディター開発効率化GitHub初心者向け
Do you need a summary in English?