Deploying your own Sweep instance via Docker

🎉 We recently changed our license to the Elastic License V2 to allow Sweep for commercial usage.

Video Guide

1. Setting up the GitHub App (5 min)

We created an app using npm to make it easier to set up the GitHub App. If you do not have it installed, install it here (opens in a new tab). If you would like ot set it up yourself, scroll down to the Manual GitHub App Setup section.

First, open a terminal and run the following commands:

git clone https://github.com/sweepai/sweep
cd sweep/self_deploy
npm install
npm start

Second, open http://localhost:3000 (opens in a new tab) in your browser, and:

1. Click the "Register GitHub App" button.
2. Click the green "Create GitHub App for USERNAME" button.
3. Pick where to install the app and then click the green "Install" button. This will take you to the app settings page.
4. Click the small "App settings" button.

Now, stop the server by pressing Ctrl+C in the terminal. Once it has stopped, run the following command:

cat .env

You should see something like this:

WEBHOOK_PROXY_URL=https://smee.io/abcdefg
APP_ID=123456
PRIVATE_KEY="-----BEGIN RSA PRIVATE KEY-----\n...\n-----END RSA PRIVATE KEY-----\n"
WEBHOOK_SECRET=213921898f9as8f92139128989f
GITHUB_CLIENT_ID=Iv1.123218f182131
GITHUB_CLIENT_SECRET=6g626312b212142132121321412

Keep this terminal open, we will need the PRIVATE_KEY and APP_ID later:

APP_ID=123456
PRIVATE_KEY="-----BEGIN RSA PRIVATE KEY-----\n...\n-----END RSA PRIVATE KEY-----\n"

2. OpenAI API Token

Please create an OpenAI API token at https://platform.openai.com/account/api-keys (opens in a new tab). You will need this later. Also, if you have access to the 32k model, please set OPENAI_DO_HAVE_32K_MODEL_ACCESS=true. If you have a beefier machine and want higher quality search, change the sentence transformers model to sentence-transformers/all-mpnet-base-v2.

OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
OPENAI_DO_HAVE_32K_MODEL_ACCESS=false
SENTENCE_TRANSFORMERS_MODEL=sentence-transformers/all-MiniLM-L6-v2

3. Deploying Sweep

Starting here it gets more involved so if you run into issues, please reach out to us on Discord (opens in a new tab) or open a GitHub issue (ironic, we know!) at https://github.com/sweepai/sweep/issues (opens in a new tab). We are online, please ping us @kevin, @william, and @luke.

As Sweep is Dockerized, you can deploy this anywhere but for this guide, we will deploy it on Digital Ocean for simplicity. If you would like to run it locally, you can follow the section in the dropdown at the bottom of this guide.

First, click the following:

1. Skip Resources Page

2. Set Environment Variables

In the Environment Variable Editor, copy and paste the environment variables from above like so:

# These two are from step 1. Rename PRIVATE_KEY to GITHUB_APP_PEM
APP_ID=1234
GITHUB_APP_PEM="-----BEGIN RSA PRIVATE KEY-----\n...\n-----END RSA PRIVATE KEY-----\n"
 
# These three are from step 2.
OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
OPENAI_DO_HAVE_32K_MODEL_ACCESS=false
SENTENCE_TRANSFORMERS_MODEL=sentence-transformers/all-MiniLM-L6-v2

Click Save to close the Bulk Editor, then click Save again to save the varaibles.

It is important that you click save twice, otherwise the variables will not be saved.

3. Info

Modify any information as needed, then click Next.

4. Review: Update Instance Resource Size

First, click Edit Plan in the Billing section.

Then, set the Instance Size.

Please select an instance with at least 2 GB RAM. The default 1 GB RAM instance will not work.

5. Create Resources!

6. Set Webhook URL in GitHub

Once the resource is created, you should get a link like this:

Navigating to this link should look like this:

1. Copy this link to your clipboard.
2. Go to your GitHub App and paste this link into the General tab Webhook URL field. Ensure SSL verification remains active.

4. Using Sweep (1 min)

Finally, you can use Sweep by creating a new issue with a title prefixed with Sweep:, like Sweep: add comments to BaseIndex.tsx and watch the magic happen!

For more details on using Sweep, see our tutorial (opens in a new tab) and advanced usage (opens in a new tab) guides.

Alternate steps:

docker pull sweepai/sweep:latest

docker run --env-file .env -p 8080:8080 sweepai/sweep:latest

Forwarding  https://4d8d8bf053be.ngrok.app -> http://localhost:8080

There's additional configurable settings like Redis caching and GPU-accelerated embeddings that we use for our hosted product that we can document further for self-hosting. If this interests you please contact us at https://discord.gg/sweep (opens in a new tab).