dumbaspl/niji-proxy repository overview

⁠niji-proxy

Proxy slitter server. Save data by only proxying the the domains that matter.

⁠Use case

You have a high quality proxy with limited bandwidth. The program using the proxy will try to proxy everything through it (images, scripts, cdn content, etc) but these requests don't actually need to be proxied since almost no CDN is doing any ip filtering. By using this project you can save bandwidth by only proxying the domains that matter (APIs, auth, etc).

⁠Limitations

• Only supports HTTP/HTTPS proxies
• Only filtering by domain (no path filtering, most trafic is encrypted anyway)

⁠Usage

⁠Configuration

Configuration is done through environment variables. (.env file is supported)

⁠NOTES

• if GOOD_PROXY has credentials, it will always use them and ignore the credentials from the client
• if GOOD_PROXY has no credentials, it will passthrough the credentials from the client
• if BAD_PROXY has no credentials, no authentication will be used (passthrough not supported to avoid leaking credentials)
• if BAD_PROXY is missing, "bad" connections will be made directly to the target host (no proxy)
• DEBUG will expose credentials in the logs, use with caution

⁠Example

LISTEN_HOST=127.0.0.1
LISTEN_PORT=8080
GOOD_HOST_REGEX=^api\.
GOOD_PROXY=http://user:pass@good-proxy-server:8080
BAD_PROXY=http://bad-proxy-server:8080

⁠Running

git clone https://github.com/dumbasPL/niji-proxy.git
cd niji-proxy
# Create .env file with the configuration

# Using docker
docker compose up

# Using node
npm install
npm start

⁠License

MIT⁠