How do you install and use the Plugsky CLI?

Key facts

What it is	Command-line client for Plugsky chat and agent sessions
Install	Follow the install instructions on the downloads page for your platform
Auth	Create an API key in your account and export it to the CLI
Models	30+ models selectable per session, from free aliases to frontier reasoning
API	OpenAI-compatible endpoint, so behavior matches the SDKs
Agent loop	Function calling and streaming are live for terminal agent workflows
Free tier	Free plan with 2 free AI models, no card required
Usage controls	Keys, RBAC and audit logs apply to CLI traffic as to any other client

TL;DR

• Install from the downloads page and authenticate with an API key.
• Pick a model per session; cheap aliases for routine work.
• Use it for chat, code questions and scripted agent loops.
• Keys and audit controls are the same as the HTTP API.
• Start on the free plan, then move to a paid tier when you need stronger models.

How it works, step by step

1. Create a Plugsky account and generate an API key in the dashboard.
2. Install the CLI using the instructions for macOS, Linux or Windows on the downloads page.
3. Export your key as an environment variable so it is never hard-coded in shell history.
4. Start a session and select a model — a fast alias for chat, a coding alias for edits.
5. Pipe files, diffs or command output into the session for context-aware answers.
6. Script non-interactive calls for CI tasks such as release notes or log triage.

Try it yourself

Open the SDK code converter →

Install and authenticate

Get the current install command for your platform from the downloads page, then create an API key in your Plugsky account. Store it as an environment variable rather than in a config file committed to a repository. The CLI talks to the same OpenAI-compatible endpoint as the SDKs, so authentication, rate limits and usage accounting behave identically. Verify the install with a one-line prompt before wiring it into scripts.

Everyday terminal workflows

• Code questions: pipe a file or stack trace into a session and ask for an explanation.
• Commit and release notes: feed a diff and request a summary in your house style.
• Log triage: stream error output and ask for likely causes, ranked.
• Scaffolding: generate a script or test skeleton, then refine it with a coding alias.
• Scripted runs: non-interactive calls in CI for repeatable text tasks.

Use fast aliases for these flows; the value is in staying in the shell, not in maximum reasoning.

Choosing a model per session

Because all 30+ models sit behind one API, switching is a flag, not a new integration. A micro or lite alias handles summaries and quick questions at low latency. A coding alias handles patches and test generation. A long-context alias answers repository-wide questions. A reasoning alias is reserved for debugging that resists cheaper attempts. Keep a small alias map in your shell config so team members use consistent defaults.

Governance and limits

CLI traffic is not exempt from controls. API keys can be scoped and revoked, RBAC governs what a key may access, and audit logs record usage for compliance. For regulated networks, point the CLI at a private Plugsky deployment — VPC, on-prem or air-gapped — instead of the public endpoint. Where the CLI is not yet a fit, call the HTTP API directly; every capability described here is available through the documented endpoints.

Honest comparison

Workflow	Plugsky CLI	Web chat UI	Direct HTTP API
Best for	Shell-centric developers	Non-technical users	Applications and services
Model switching	Per-session flag	Dropdown	Request parameter
Scriptable	Yes, non-interactive mode	No	Fully
Governance	Same keys, RBAC and audit	Same account controls	Same account controls
Offline networks	With private deployment	With private deployment	With private deployment

Frequently asked questions

Is the Plugsky CLI free to use?

The CLI is a client for the API, so usage is governed by your plan. The free plan includes 2 free AI models with no card, and paid self-serve plans are flat monthly — see the live pricing page.

Where do I get the install command?

From the downloads page, which documents the current install method for each supported platform. Install steps are versioned there rather than in this article.

Which models can the CLI use?

All models available to your account — 30+ behind one API — selected per session. Model-level capability, including tool support, is documented in the catalogue.

Can I use the CLI in CI pipelines?

Yes, through non-interactive invocations. Keep the API key in your CI secret store and scope the key to the least access the job needs.

Does the CLI keep my code on my machine?

The CLI sends only the context you provide to the API endpoint you configure. For stricter isolation, point it at a private Plugsky deployment.

How is usage tracked?

The same usage accounting and admin controls apply as for any API client, including audit logs and per-key visibility.