Getting started

Requirements

Spark currently requires Node.js >=26 <27. The published product contains the CLI dispatcher, native TUI, daemon, and Cockpit host.

Install

The managed installation is recommended because it supports atomic upgrades and rollback:

pnpm dlx @zendev-lab/spark install --managed
spark version --json
spark update status --json

You can instead keep the package manager in charge of the installation:

npm install --global @zendev-lab/spark
spark --help

Package-manager and source-checkout installations report update instructions but never replace themselves.

Run the health check before troubleshooting a host:

spark doctor

Configure a model

Open the interactive TUI:

spark

Use /login to inspect available provider authentication and start the provider’s interactive login flow. Use /model to inspect or select the active model. When Spark prompts for an API key, enter it in the prompt; do not put secrets in project files, config.json, or shell history.

Complete a first run

For a foreground, non-interactive answer:

spark run "Summarize this repository and identify its validation command."

Use JSON mode for scripts:

spark run --json "List the top-level packages."

For an interactive session, stay in spark or run:

spark tui "Inspect the current project before proposing a change."

Spark starts or contacts the local daemon as needed. Run spark daemon status --json to inspect the service rather than guessing from frontend behavior.

Next steps

• Plan and implement your first change.
• Browse the complete feature map without learning every command.
• Learn the TUI’s progressive controls.
• Choose between foreground runs, background work, and sessions.
• Open the Cockpit web surface.
• Use automation only when ordinary Plan and Implement are not enough.