PluelyPluelyDocs

Settings & Configuration

Connect a Local CLI

The Connect your CLI area of the Dev Space (dashboard → Dev space) lets Pluely detect and use an AI CLI you already have installed on your machine, so your requests run through that local tool instead of one of Pluely's managed models.

Supported CLIs

Pluely can detect and connect to five popular AI CLIs — each has its own setup page:

  • Claude Code — Anthropic's CLI; Sonnet / Opus / Haiku on your Claude subscription.
  • Gemini CLI — Google's CLI with a generous free tier; Gemini 3 Pro and 2.5 Pro/Flash.
  • OpenAI Codex — GPT-5.1-class models on your ChatGPT plan.
  • Qwen Code — free Qwen3-Coder models.
  • Ollama — fully local, fully offline models.

If any of these are installed locally, Connect your CLI will find them without you needing to type a path or configuration by hand.

Connecting a CLI

  1. Open the Dev Space and go to Connect your CLI.
  2. Pluely scans your machine for supported, locally-installed CLIs (versions shown).
  3. Click Connect on the one you want.
  4. Your connected CLI now appears at the top of the overlay's "Select a model" picker as Your CLI, ready for any question.

Choosing a model — for every CLI

Every connected CLI gets a Model panel directly under its row:

  • Claude Code, Gemini CLI, OpenAI Codex, Qwen Code — pick from the well-known models (e.g. Sonnet/Opus/Haiku, Gemini 3 Pro, GPT-5.1 Codex, Qwen3 Coder), type a custom model id, or leave Default (CLI decides).
  • Ollama — the panel lists your locally pulled models; picking one is required (there's no default).

Your selection shows on the connection card and in the overlay picker (e.g. Claude Code · opus, Ollama · llama3.2).

Screenshots and images

The Model panel tells you whether the current setup can see images. Claude Code, Gemini CLI, OpenAI Codex, and Qwen Code are vision-capable; Ollama depends on the pulled model (llava, gemma3, qwen-vl, and other multimodal families enable image support automatically). When vision is available, Capture, Selection, and "Use image" all work through your CLI.

Thinking models

Reasoning models are fully supported on every CLI: their chain-of-thought (thinking) output is stripped automatically, so you always get a clean final answer.

Refresh and Disconnect

  • Refresh — the refresh button beside the "Connect your CLI" heading re-scans, e.g. right after installing a new CLI (it spins while detecting).
  • Disconnect — remove the current CLI connection. The overlay's "Your CLI" group goes away until you connect again.

This is a Pro feature

Connecting a CLI requires an active license — any paid plan or the Dev Pro lifetime license. Without one, this area shows an "Upgrade to unlock CLI" prompt instead of the connect flow.

Why connect a CLI

Running your requests through a CLI already installed on your machine — Claude Code, Gemini CLI, OpenAI Codex, Qwen Code, or Ollama — keeps those requests off Pluely's servers entirely. The request is handled locally by the tool you connected, and Pluely simply routes to it. This is the option to reach for when you want the strongest guarantee that your questions and context never leave your machine.

Last updated 2026-07-10 · pluely.com