Command Reference

The BroomVA CLI provides six command groups: auth, prompts, skills, context, config, and daemon. All commands are implemented in packages/cli/src/commands/ as Commander.js subcommands.

Global options

These options are available on all commands:

Flag	Description
`--api-base <url>`	Override the API base URL
`--token <token>`	Override the auth token
`--no-color`	Disable colored output
`-V, --version`	Print the CLI version
`-h, --help`	Print help information

`broomva auth`

Manage authentication with the BroomVA platform. See Authentication for detailed usage.

Subcommand	Description
`auth login`	Authenticate using device-code flow (RFC 8628)
`auth login --manual`	Authenticate using manual token copy-paste
`auth logout`	Remove stored credentials from `~/.broomva/config.json`
`auth status`	Show current auth status (validates token against server)
`auth status --json`	Output auth status as JSON
`auth token`	Print the raw token string (for use in scripts)

`broomva prompts`

Manage prompts stored on the platform. Prompts are versioned text templates that can be used in chat conversations, API calls, and agent configurations.

Subcommand	Description
`prompts list`	List all prompts accessible to your account
`prompts list --category <cat>`	Filter by category
`prompts list --tag <tag>`	Filter by tag
`prompts list --model <model>`	Filter by target model
`prompts list --mine`	Show only your prompts (requires auth)
`prompts list --json`	Output as JSON
`prompts get <slug>`	Fetch a specific prompt by slug
`prompts get <slug> --raw`	Output only the prompt content (no metadata)
`prompts get <slug> --json`	Output as JSON
`prompts create`	Create a prompt (see options below)
`prompts update <slug>`	Update an existing prompt
`prompts delete <slug>`	Delete a prompt
`prompts pull <slug>`	Download a prompt to a local markdown file with frontmatter
`prompts push <file>`	Upload a local markdown file with frontmatter to the platform

Creating a prompt

broomva prompts create \
  --title "Code Review Assistant" \
  --content @prompt.md \
  --summary "Reviews code for bugs, style, and performance" \
  --category "development" \
  --model "claude-sonnet-4-20250514" \
  --tags "code-review,development" \
  --visibility public

The --content flag accepts either a literal string or @file to read from a file.

Pull/push workflow

The pull and push commands enable a local-first workflow for prompt management:

# Download a prompt to a local file
broomva prompts pull system-default
# Pulled "System Default" → system-default.md

# Edit the local file...
# The file has YAML frontmatter with title, slug, category, tags, etc.

# Upload changes back to the platform
broomva prompts push system-default.md
# Updated prompt: system-default

The frontmatter format:

---
title: "System Default"
slug: "system-default"
summary: "Default system prompt"
category: "system"
model: "claude-sonnet-4-20250514"
tags: ["system", "default"]
visibility: "private"
---
Your prompt content here...

When pushing with --create, a new prompt is created. Without the flag, the slug from the frontmatter is used to update an existing prompt.

`broomva skills`

Browse and install skills -- packaged capabilities that can be installed into agent runtimes. Skills are discovered from SKILL.md files in repositories and published to the skills marketplace.

Subcommand	Description
`skills list`	List available skills grouped by layer
`skills list --layer <id>`	Filter by layer ID
`skills list --json`	Output as JSON
`skills get <slug>`	Get details for a skill (name, description, install command, URL)
`skills get <slug> --json`	Output as JSON
`skills install <slug>`	Install a skill (delegates to `npx skills add broomva/<slug>`)

Example

# List available skills
broomva skills list
# Shows skills grouped by layer with slug, name, and description

# Get details for a specific skill
broomva skills get deep-dive-research
# deep-dive-research
# Deep research skill with multi-source synthesis
# Layer: research
# Install: npx skills add broomva/deep-dive-research
# URL: https://broomva.tech/skills/deep-dive-research

# Install a skill
broomva skills install deep-dive-research
# Running: npx skills add broomva/deep-dive-research
# Installed skill: deep-dive-research

The install command fetches the skill's install command from the API and executes it via execSync. If the API is unreachable, it falls back to npx skills add broomva/<slug>.

`broomva context`

Show the project context -- the conventions, stack, and features that inform the platform's behavior.

Subcommand	Description
`context show`	Display the full context (app info, conventions, stack, features)
`context show --json`	Output as JSON
`context conventions`	Show conventions only
`context conventions --json`	Output as JSON
`context stack`	Show tech stack only
`context stack --json`	Output as JSON

Example

# Show what conventions the platform follows
broomva context conventions
# packageManager: bun
# linter: biome
# auth: better-auth
# orm: drizzle
# ...

# Show the tech stack
broomva context stack
# framework: Next.js 16
# runtime: Bun
# database: PostgreSQL
# ...

The context command queries the platform API for the project context. This is useful for AI agents that need to understand the platform's conventions before generating code.

`broomva config`

Manage CLI configuration stored in ~/.broomva/config.json.

Subcommand	Description
`config set <key> <value>`	Set a configuration value
`config get [key]`	Get a configuration value (or show all if no key given)
`config get --json`	Output as JSON
`config reset`	Reset configuration to defaults

Configurable keys

Key	Description	Default
`apiBase`	API base URL	`https://broomva.tech`
`defaultFormat`	Default output format	--
`daemon.heartbeatIntervalMs`	Daemon heartbeat interval	30000
`daemon.dashboardPort`	Daemon dashboard port	4200
`daemon.symphonyUrl`	Symphony service URL	--
`daemon.arcanUrl`	Arcan service URL	--
`daemon.lagoUrl`	Lago service URL	--
`daemon.autonomicUrl`	Autonomic service URL	--
`daemon.incidentThreshold`	Consecutive failures before incident	3

Example

# Point the CLI at a local development server
broomva config set apiBase http://localhost:3001

# Configure daemon to check services faster
broomva config set daemon.heartbeatIntervalMs 10000

# Reset to defaults
broomva config reset

`broomva daemon`

Manage the broomvad runtime daemon -- a long-running process that monitors platform and Life service health, reports incidents, and provides a local dashboard.

Subcommand	Description
`daemon start`	Start the broomvad daemon
`daemon start --env local`	Start in local mode (uses localhost URLs)
`daemon start --env railway`	Start in railway mode (default, uses configured URLs)
`daemon start --port <port>`	Override dashboard port
`daemon start --interval <ms>`	Override heartbeat interval
`daemon stop`	Stop the running daemon
`daemon status`	Check if the daemon is running and show sensor states
`daemon status --json`	Output as JSON
`daemon logs`	View daemon log entries
`daemon logs --lines <n>`	Show last N log entries (default: 20)
`daemon logs --level <level>`	Filter by level (debug, info, warn, error)
`daemon logs --json`	Output as JSON
`daemon tasks`	Show active incidents
`daemon tasks --all`	Include resolved incidents
`daemon tasks --json`	Output as JSON

Health sensors

The daemon registers multiple health sensors that run on each heartbeat tick:

Sensor	Description
`SiteHealthSensor`	Checks the main website (broomva.tech) availability
`ApiHealthSensor`	Checks the API endpoint health
Railway sensors	Checks Railway-deployed services (Symphony, Arcan, Lago, Autonomic)

Each sensor reports a status of healthy, degraded, or down with an optional latency measurement. When a sensor reports down for consecutive ticks exceeding the incidentThreshold, an incident is opened automatically.

Local mode

In local mode (--env local), the daemon uses localhost URLs for all services:

broomva daemon start --env local
# Starting broomvad (env: local, api: http://localhost:3000)
# Dashboard at http://localhost:4200

Service	Local URL
Symphony	`http://localhost:8080`
Arcan	`http://localhost:8081`
Lago	`http://localhost:8082`
Autonomic	`http://localhost:8083`

Dashboard

The daemon serves a local HTTP dashboard showing real-time sensor states, active incidents, and Symphony connection status.

Example

# Start the daemon
broomva daemon start
# Starting broomvad (env: railway, api: https://broomva.tech)
# Dashboard at http://localhost:4200
# broomvad running (PID 12345, env: railway). Press Ctrl+C to stop.

# Check status from another terminal
broomva daemon status
# Daemon running (PID 12345)
# Started: 2026-03-22T10:00:00Z
# Last tick: 2026-03-22T10:01:30Z (#3)
# Symphony: connected
#
# Sensors:
#   ● site-health: OK (120ms)
#   ● api-health: OK (85ms)
#   ● railway-arcan: OK (200ms)

# View recent logs
broomva daemon logs --lines 5

# Stop the daemon
broomva daemon stop
# Daemon stopped (PID 12345)

The daemon feature connects to the Life Agent OS runtime and the Symphony orchestration service. For full local inference without the daemon, see the Arcan documentation for running the Rust runtime directly.

Command Reference

On this page