Vela

An AI coding agent for your terminal — reads files, writes code, and runs commands.

Install

Install the latest published release from npm:

npm install -g @vela-labs/agent

Setup

Run vela, choose OpenRouter or OpenAI, and follow the secure in-terminal setup. Vela stores provider-scoped keys in ~/.vela/config.json with user-only permissions.

For OpenAI, you can paste an existing Platform API key or use the guided ChatGPT-account path to open the API keys page. ChatGPT subscriptions and OpenAI API billing are separate.

Usage

vela

That's it. Type your task and press Enter.

Keyboard shortcuts:

• Esc — cancel the current response
• Ctrl+C — quit
• PageUp / PageDown — scroll the transcript

Commands:

• /provider — switch between OpenRouter and OpenAI
• /login — update the active provider's API key
• /model — choose a model from the active provider
• /thinking — change reasoning effort
• /new — start a new session

What it can do

Vela has access to these tools in every session:

| Tool | What it does | |---|---| | bash | Run shell commands | | read_files | Read one or more files (with line ranges) | | write_file | Create new files | | edit_file | Edit existing files with targeted find-and-replace |

Providers

OpenRouter defaults to DeepSeek V4 Flash. OpenAI defaults to GPT-5.5. Each provider keeps its own selected model and reasoning preference.