Tools And Setup

Table of Contents

• TL;DR
• Overview
  • When to use what
• GitHub Copilot CLI
• OpenCode (experimental)
• OpenSpec
• Other CLI tools
• Installation and updates
• Authentication
• GitHub Copilot subscription policy
  • Personal use

TL;DR

Get up and running:

sjust sparkdock-upgrade    # install all tools

Start coding with AI:

Authenticate (one-time):

All GitHub-based tools authenticate against github.com using your SparkFabrik organization account.

The rest of this page covers how each tool works, when to use which, and the subscription policy.

Overview

All AI development tools are installed and configured by sparkdock, our workstation provisioning system. You should not need to install anything manually. Run sjust sparkdock-upgrade to provision or update everything (sjust is sparkdock's just-based task runner).

To install or update a specific tool, use tags:

sjust sparkdock-install-tags copilot-cli  # just Copilot CLI
sjust sparkdock-install-tags opencode     # just OpenCode
sjust sparkdock-install-tags skills       # just shared skills
sjust sparkdock-install-tags npm_packages # just npm packages (OpenSpec, etc.)

When to use what

AI coding tools operate in two modes: reactive (suggest code, wait for you to apply it) and agentic (plan, act, observe, adjust in a loop). Both are useful. The agentic mode (where the AI reads your codebase, forms a plan, executes it, runs tests, and course-corrects) is where the biggest productivity shift is happening, and it lives primarily in the terminal.

CLI agents have full terminal access, so they can run tests, check git status, install dependencies, call APIs. IDE agent mode is agentic too but sandboxed within the editor. Use the IDE for inline edits and code review; reach for a CLI agent when the task involves implementation, debugging, or anything that benefits from an autonomous work loop.

GitHub Copilot CLI

GitHub Copilot CLI is our primary AI coding assistant in the terminal. Sparkdock installs it and configures shell aliases for quick access to different models.

One-shot mode (co)

Send a prompt and get a single response, good for quick questions, code generation, and explanations.

co "explain what this function does" < src/auth/session.ts
co "write unit tests for the rate limiter" < src/middleware/rate-limit.ts

You can pipe files into co via stdin or let it read your codebase from the current directory. The default model is configured by sparkdock; you can override it per-invocation with --model:

co --model claude-sonnet-4.5 "review this for security issues" < src/auth/token.ts

Interactive mode (ico)

Starts a persistent chat session, good for iterative work, debugging, and multi-step tasks where you want to build on previous context.

ico                    # start an interactive session
ico --model gemini-3-pro  # start with a specific model

Inside an interactive session you can use slash commands (/help to list them), switch models during the conversation, and reference previous responses.

IDE integration

GitHub Copilot also runs as an extension in VS Code and JetBrains IDEs, providing inline completions and a chat panel. Install it from your IDE's extension marketplace and sign in with your GitHub account.

Skills and slash commands work in both the terminal CLI and the IDE extension. A skill is a set of instructions that teaches the AI how to perform a specific task (e.g., how to use glab, how to follow project conventions); a slash command triggers a specific workflow. Project-level skills go in .github/skills/ and prompts in .github/prompts/.

OpenCode (experimental)

OpenCode is an open-source, model-agnostic terminal coding agent. Like Copilot CLI, it runs the full agentic loop: reading your codebase, planning changes, executing them, running tests, and adjusting when things fail. It is currently experimental at SparkFabrik; we are evaluating it alongside Copilot CLI.

Start a session:

c                          # start in the current directory
c --agent the-architect    # start with a specific agent profile

OpenCode uses the current directory as project context. Press Tab to switch agent profiles, type / for slash commands (including /opsx:* for OpenSpec workflows).

What sets it apart:

• Model-agnostic: works with Claude, GPT, Gemini, and others; no vendor lock-in
• Full tool use: reads, edits, creates files; runs shell commands; manages git
• Skills and agent profiles: loads from ~/.agents/skills/ (system), .opencode/skills/ (project), and ~/.config/opencode/agents/ (agent profiles)
• Permission controls: sparkdock configures rules that require confirmation before destructive operations (force-push, file deletion outside project, etc.)

Configuration: sparkdock installs the config at ~/.config/opencode/opencode.json. You generally don't need to edit it.

OpenSpec

OpenSpec is a framework for capturing requirements, design decisions, and implementation tasks as structured artifacts before and during AI-assisted development. It's installed as an npm package by sparkdock. Verify with:

openspec --version

OpenSpec is covered in detail on its own page: Spec-Driven Development.

Other CLI tools

Sparkdock also installs CLI tools that aren't AI-specific but that the coding agents use via skills:

Both tools work standalone in your terminal and are also used by Copilot and OpenCode through skills (e.g., the glab skill lets the AI fetch issue details, read MR discussions, and check pipelines on your behalf).

Installation and updates

Everything is managed by sparkdock. Here are the commands you'll use most:

If a tool is missing or outdated, sjust sparkdock-upgrade is always the safe default.

What sparkdock installs

Sparkdock also configures shell aliases, zsh completions, and OpenCode permissions.

Authentication

All GitHub-based tools authenticate against github.com using your SparkFabrik organization account via OAuth. This is safe and expected. No API keys to manage manually, no third-party services. Your credentials are stored locally (in your OS keychain for Copilot CLI, in OpenCode's auth store for OpenCode).

GitHub Copilot CLI

Copilot CLI has its own authentication, separate from the gh CLI. Run copilot login from your terminal (or /login inside an active session):

copilot login

This opens your browser at github.com/login/device where you enter the one-time code displayed in the terminal. Select GitHub.com, authorize, and you're set. The token is stored in your OS keychain under the service name copilot-cli.

If you've already authenticated with gh auth login, Copilot CLI can use that token as a fallback. But we recommend running copilot login explicitly so Copilot CLI has its own stored credential.

For the full reference, see the official docs: Authenticating GitHub Copilot CLI.

GitHub Copilot in VS Code / JetBrains

Sign in through the Copilot extension sidebar. It opens the same GitHub OAuth flow in your browser. Once authorized, the extension stays authenticated.

OpenCode (experimental)

Open an OpenCode session (c), then run the /connect command:

/connect

Select GitHub Copilot from the provider list. OpenCode gives you a device code and a link to github.com/login/device. Enter the code, authorize, done.

For the full reference, see the OpenCode docs: GitHub Copilot provider.

glab (GitLab CLI)

glab authenticates against GitLab, not GitHub. Run:

sjust gitlab-configure-glab

This is a one-time setup handled by sparkdock.

gh (GitHub CLI)

gh auth login

Select GitHub.com, authenticate via browser. This is also the fallback credential for Copilot CLI if copilot login hasn't been run.

GitHub Copilot subscription policy

Eligibility

All employees with access to GitHub can use GitHub Copilot. You should receive a subscription during onboarding.

Usage policy

1. Employees are encouraged to use AI suggestions but must use their own judgment when applying them. Always review suggested code and understand it before implementing it. The name is Copilot, not Autopilot.

2. Employees must comply with all company policies when using GitHub Copilot. Non-compliance is the employee's responsibility.

3. The service is regulated by the GitHub Terms for Additional Products and Features and the GitHub Privacy Statement. Code snippets are transmitted in real-time for suggestions but are not retained by default. For details on data handling, see the GitHub Copilot Trust Center. It is the employer's responsibility to verify that these terms are acceptable for their specific use case; if in doubt, ask your team lead or manager.

4. This policy is reviewed periodically and updated as needed.

Personal use

GitHub attaches the Copilot subscription to your personal GitHub account. When the organization seat is assigned, any existing personal subscription is automatically cancelled. We recognize this means the tools you use for work are the same tools available on your personal projects.

Here is what is and isn't covered:

• Open source and community contributions: encouraged. Contributing to open source projects, building community tools, and participating in hackathons all build skills that benefit you and the company.
• Personal non-commercial projects: allowed. Learning new technologies, hobby projects, and personal tools are fine, within the guidelines below.
• Personal commercial projects: not covered. Freelance work, side businesses, SaaS products, and paid consulting deliverables should not use the company-provided subscription.
• Projects that compete with the company: not allowed. Personal projects must not compete with SparkFabrik's products, services, or business areas.

A few guidelines:

1. Work takes priority. Premium requests are metered and have a cost. If your usage is unusually high, we may ask about it.
2. Keep company and personal work separate. Personal projects must not include, reuse, or be derived from company code, proprietary logic, or confidential information.
3. Your personal projects are yours. The company makes no intellectual property claim on code you write for personal or open source projects, provided it is developed outside working hours and is not related to the company's business.
4. This is trust-based. We'd rather give clear boundaries than build surveillance. If costs become unsustainable, we'll revisit the policy with transparency.

If you're unsure whether a specific use case qualifies, ask your manager.

Requesting a subscription

If you weren't granted a subscription during onboarding:

1. On the GitHub Copilot settings page, locate Get Copilot from an organization
2. Use Ask admin for access on the SparkFabrik organization
3. Confirm with an HR representative via the #support-hr channel

The subscription is through our GitHub Organization. Any existing personal Copilot subscription on the same account will be automatically cancelled and refunded by GitHub.

Refunding a previously purchased subscription

If you already purchased a personal GitHub Copilot subscription, GitHub will refund the unused portion when the Organization subscription is attached. You can request a refund for the partial paid subscription using the company's employee refund platform, only if it was previously approved as a company expense.