Skip to main content
The NOFire AI CLI is a terminal chat interface for investigating incidents, querying production, and running diagnostics directly from your terminal. It connects to the same NOFire AI backend as Slack and the web dashboard, with one key addition: it can run read-only commands on your local machine (kubectl, aws, git, helm) and feed the results back into the conversation.
Conversations are persisted and accessible at my.nofire.ai, so you can pick up where you left off in any interface.

Authenticate

Get an API token from my.nofire.ai/dashboard/api-keys (see API Keys), then:
Paste your token when prompted. Credentials are stored in your OS keychain (macOS Keychain, GNOME Keyring, Windows Credential Manager). Verify everything works:
This checks your config, auth, API connectivity, and which local tools are available.

Usage

Interactive chat

Opens a full-screen terminal UI with streaming responses, markdown rendering, and local tool execution. This is the primary way to use the CLI. Continue an existing thread:
Focus on a specific service:

Single question

Sends one message, streams the response, and exits. Useful for scripting or quick checks.

Keyboard Controls

The TUI has two modes, toggled with Tab: Input mode (default) Scroll mode (press Tab to enter)

Local Tool Execution

When investigating an issue, NOFire AI can request to run commands on your machine — pulling live data from your infrastructure without you switching terminals. Supported tools:
All commands are read-only. The CLI blocks any write, delete, or destructive operations. Commands that don’t match the read-only allowlist are rejected.

Commands Reference