Configure Cursor to use AWS Bedrock

If you’re subscribed to Cursor Pro, you can configure Cursor to use AWS Bedrock. However, doing so requires creating an IAM user for Cursor in your AWS account. Unfortunately, for security, many employers disallow creating IAM users.

Fortunately, there’s a way to circumvent this problem. In addition to AWS Bedrock, Cursor can configured to use, OpenAI, Anthrophic, Google or any provider that offers an OpenPI compatible compatible API.

Luckily, there are many proxies that

Offer an OpenAI compatible API
Proxy requests to AWS BedRock

LiteLLM is one.

First, buy an inexpensive Cloud VM (This is necessary as Cursor refuses to use any proxy running locally).

Install LLM

pip install 'litellm[proxy]'

Add a default AWS profile in ~/.aws/config

mkdir -p ~/.aws
vim ~/.aws/config

Create a LiteLLM config file ~/.config/litellm/config.yaml (Replace your-master-key with an unguessable string)

mkdir -p ~/.config/litellm

general_settings:
  master_key: your-master-key

model_list:
  - model_name: ll-op-45
    litellm_params:
      model: bedrock/us.anthropic.claude-opus-4-5-20251101-v1:0
      aws_profile_name: default
      aws_region_name: us-east-1

Start LiteLLM

litellm --config config.yaml

Then,

Open Cursor’s Command Palette, search for “Cursor Settings: Models” and open the “Models” section”
Change OpenAI Base URL to http://1.2.3.4:4000 (Replace 1.2.3.4 with your VM’s public IP)
Change OpenAI API Key to your-master-key
Lastly, click on “Add Custom Model” link and add ll-op-45

That's it! Choose ll-op-45 as your model and start chatting with your Agent.

Tunneling (Optional)

If you

Prefer to start LiteLLM locally and not on a Cloud VM
Or have a Cloud VM that doesn’t have a public IP address

install Cloudflare Tunnel or ngrok or a different alternative so LiteLLM becomes available at a public domain.

Daemonize LiteLLM (Optional)

If you’d like to daemonize LiteLLM, add a system service file

vim /etc/systemd/system/litellm.service

[Unit]
Description=LiteLLM
After=network.target

[Service]
Type=simple
User=ubuntu
WorkingDirectory=/home/ubuntu
ExecStart=/home/ubuntu/.local/bin/litellm --config /home/ubuntu/.config/litellm/config.yaml
Restart=always
RestartSec=3

[Install]
WantedBy=multi-user.target

and start it

sudo systemctl daemon-reload
sudo systemctl start litellm

Run

sudo systemctl status litellm

to verify if LiteLLM started successfully

Run

sudo journalctl -u litellm -f

to stream logs

If you’d like to start LiteLLM automatically when your VM starts or restarts, run

sudo systemctl enable litellm

Configure Cursor to use AWS Bedrock

Tunneling (Optional)

Daemonize LiteLLM (Optional)

Resources