omp with AI Gateway

omp (oh-my-pi) is an open-source terminal coding agent forked from Pi. It inherits Pi's first-class vercel-ai-gateway provider, so it already knows the gateway's base URL and fills its model picker from the gateway catalog. Unlike Pi, omp reads the credential from the environment, so the whole setup is one variable.

Run the Vercel CLI setup command for omp:

npx vercel ai-gateway setup --agent omp

The command provisions a key and adds the export to your shell profile, backed by the macOS Keychain where available.

If you can't use the Vercel CLI, configure omp manually:

1. Install omp

   Install omp with its official installer:

   Terminal

   curl -fsSL https://raw.githubusercontent.com/can1357/oh-my-pi/main/scripts/install.sh | sh

2. Create an API key

   Go to the AI Gateway section in the Vercel dashboard sidebar and click API keys to create a new API key.

3. Export the key

   omp authenticates its vercel-ai-gateway provider from the AI_GATEWAY_API_KEY environment variable. Add the export to your shell profile:

   ~/.zshrc

   export AI_GATEWAY_API_KEY="your-ai-gateway-api-key"

   Open a new terminal so the variable is loaded.

   omp also reads .env files, including ~/.omp/.env, so the export can live there instead of your shell profile. Its own /login flow stores credentials in a local database. The environment variable has the same effect without it.

4. Pick a model

   Start omp and choose a gateway model with /model, or name one when you launch:

   Terminal

   omp --model vercel-ai-gateway/anthropic/claude-opus-5

   Unlike Pi, which takes bare gateway model IDs, omp prefixes them with the provider id: vercel-ai-gateway/ followed by the gateway's creator/model-name format. To browse the catalog from the terminal, with context sizes and capabilities:

   Terminal

   omp models vercel-ai-gateway

• Browse the models catalog for available models
• Read omp's own provider documentation
• Configure spend limits on your API key
• Connect your other agents with the Vercel CLI