nexotao-cli

Interactive terminal access to every live Nexotao AI model using one API key.

Install

Requires Node.js 20 or newer.

npm install --global nexotao-cli
nexotao-cli login
nexotao-cli

Authentication

nexotao-cli login prompts for the API key without echoing it, validates it against Nexotao, then writes it to ~/.config/nexotao/config.json with file mode 0600. Authentication is resolved in this order:

1. NEXOTAO_API_KEY
2. the saved config
3. --api-key for an explicit one-off test

Do not place --api-key in shell history for normal use. Set NEXOTAO_BASE_URL or pass --base-url only when testing another Nexotao API deployment.

Commands

nexotao-cli                         # interactive chat
nexotao-cli login                   # validate and save an API key
nexotao-cli logout                  # remove the saved API key
nexotao-cli models list             # fetch the live model catalog
nexotao-cli models select           # interactively save a default model
nexotao-cli models select gpt-5-mini
nexotao-cli chat "Explain RAG briefly"
nexotao-cli chat -m claude-opus-4-8 "Review this idea"
nexotao-cli doctor
nexotao-cli --json doctor
nexotao-cli --json request get /v1/models

Inside interactive chat:

• /model switches models and saves the new default.
• /clear clears conversation context.
• /help shows interactive commands.
• /exit exits.

Prompts can also be piped:

printf 'Summarize this in one line' | nexotao-cli chat -m gpt-5-mini

JSON contract

Use the global --json option for automation. JSON mode writes exactly one JSON document to stdout and does not stream chat deltas.

Successful commands use this envelope:

{ "ok": true, "data": { "models": [], "selected": null } }

Errors exit non-zero and use this credential-safe envelope:

{
  "ok": false,
  "error": {
    "code": "auth.api_key_missing",
    "message": "No API key found. Run: nexotao-cli login",
    "status": 401
  }
}

Command-family examples:

nexotao-cli --json models list
nexotao-cli --json chat -m gpt-5-mini "Say hello"
nexotao-cli --json request get /v1/models

request get is intentionally read-only. It applies configured authentication, client headers, API error handling, and output redaction. Add a named high-level command before adding recurring workflows to the raw escape hatch.

Development and publishing

npm install
npm run check
npm pack --dry-run
npm publish

Publishing requires npm access to the nexotao-cli package. prepublishOnly runs formatting, type checks, tests, and a production build.