Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/MiscellaneousStuff/anterion
Open-source software engineer
https://github.com/MiscellaneousStuff/anterion
agent llm programming
Last synced: 2 months ago
JSON representation
Open-source software engineer
- Host: GitHub
- URL: https://github.com/MiscellaneousStuff/anterion
- Owner: MiscellaneousStuff
- License: mit
- Created: 2024-04-10T03:44:02.000Z (9 months ago)
- Default Branch: main
- Last Pushed: 2024-05-12T15:15:39.000Z (8 months ago)
- Last Synced: 2024-08-03T14:06:53.806Z (5 months ago)
- Topics: agent, llm, programming
- Language: Python
- Homepage: http://anterion.ai
- Size: 2.38 MB
- Stars: 166
- Watchers: 8
- Forks: 22
- Open Issues: 5
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- awesome-devins - GitHub
README
# Anterion Web Beta
We've recently released a new web version of Anterion that requires no setup or LLM API key to use! You can access it [here](https://beta.anterion.ai/)!:
[](https://discord.gg/nbY6njCuxh)
# 🤖 Anterion Agent
## 📖 What is Anterion?
Anterion is an open-source AI software engineer.
Anterion extends the capabilities of `SWE-agent` to plan and execute open-ended engineering tasks, with a frontend inspired by
`OpenDevin`.
We've equiped Anterion with easy deployment and UI to allow you to fix bugs and prototype ideas at ease.
## 🎥 Install Tutorial (Windows/WSL)
## 🏁 Getting Started
🎉 Get on board with Anterion by doing the following! 🎉
### Prerequisites
* Linux, Mac OS, or [WSL on Windows](https://learn.microsoft.com/en-us/windows/wsl/install)
* [Docker](https://docs.docker.com/engine/install/)
* [Python](https://www.python.org/downloads/) >= 3.11
* [NodeJS](https://nodejs.org/en/download/package-manager) >= 18.17.1
* [Miniconda](https://docs.anaconda.com/free/miniconda/miniconda-install/)
You will need to setup all three components of the system before being able to run it:
### 1. `OpenDevin` Setup
Before setting up OpenDevin, make a new conda environment and activate
it by doing the following:
```bash
conda create --name anterion python=3.11
conda activate anterion
```
To setup OpenDevin, run the following command in the `anterion` directory:
```bash
make build-open-devin
```
### 2. `SWE-agent` Setup
Next you will need to setup the `SWE-agent`.
To start, you will need to `cd` to the `SWE-agent` directory, and run the following
command:
```bash
cd SWE-agent
conda env create -f environment.yml
conda activate swe-agent
```
You will need to create a file called `keys.cfg` inside of the `SWE-agent`
directory:
```bash
OPENAI_API_KEY: ''
ANTHROPIC_API_KEY: ''
GITHUB_TOKEN: ''
```
And add the following
`.env` file inside of the `SWE-agent` directory:
```bash
NETLIFY_AUTH_TOKEN=""
NETLIFY_SITE_ID=""
```
Netlify deployments are optional. If you do not want to use them or don't have netlify installed, you can leave both fields as empty strings.
From the `SWE-agent` directory head back to the `anterion` directory and run the following command to setup `SWE-agent`
```bash
cd ..
make build-swe-agent
```
### 3. `microservice` Setup
Finally, you need to setup the `microservice`, which ties together the
`OpenDevin` frontend and the `SWE-agent` agent.
First, within the `microservice` directory, create a new
directory called `docker_volume` which will be used to store files.
```bash
cd ./microservice
mkdir docker_volume
```
Then you need to create a `.env` file in the `microservice` directory
like the following:
```bash
OPENAI_API_KEY=
ANTHROPIC_API_KEY=
SWE_AGENT_PATH=
PYTHON_PATH=
DOCKER_HOST_VOLUME_PATH=
DOCKER_CONTAINER_VOLUME_PATH=/usr/app
SWE_AGENT_PER_INSTANCE_COST_LIMIT=
SWE_AGENT_TIMEOUT=25
SWE_AGENT_MODEL_NAME=gpt4
```
#### Ollama Support
If you want to use an `Ollama` model, change `SWE_AGENT_MODEL_NAME` to look like the following:
```bash
SWE_AGENT_MODEL_NAME=ollama:
```
For example, if you want to try the new LLama 3 model, use the following line:
```bash
SWE_AGENT_MODEL_NAME=ollama:llama3
```
Next, head from the `microservice` directory `cd` to the `anterion` directory and return to the `anterion` environment using:
```bash
cd ..
conda deactivate
```
Finally, run the following command from the `anterion` directory to build the microservice:
```bash
make build-microservice
```
### Usage
To now run Anterion, you need to be in the `anterion` environment.
Then you need to run the frontend and the backend.
Run the following command from the `anterion` directory to run both together:
```bash
./run.sh
```
You may have to change permissions for the file first:
```bash
chmod +x run.sh
```
If that isn't working for some reason, run both of them separately:
```bash
make run-frontend
```
```bash
make run-backend
```
## 🙏 Special Thanks!
We'd like to say thanks to these amazing repos for inspiration!
- [OpenDevin](https://github.com/OpenDevin/OpenDevin)
- [SWE-agent](https://github.com/princeton-nlp/SWE-agent)