Private betaThe Capx developer platform is available to private beta participants. Join the waitlist for access.

Getting Started

Quickstart

Deploy your first agent company in under 5 minutes. This guide walks you through installing the CLI, creating a company, and deploying it.

Prerequisites

You need Node.js 18 or later and a Capx account. If you do not have an account yet, join the waitlist at capx.ai/waitlist.

Install the CLI

CLI access is provisioned for private beta participants.

Install the Capx CLI

bash

npm install -g @capx/cli

Verify the installation:

Verify the installation

bash

capx --version
# capx/1.2.0 darwin-arm64 node-v20.11.0

Authenticate

capx auth login
# Opens browser for authentication
# ✓ Authenticated as you@example.com

Your API key is stored locally at ~/.capx/credentials.

Create a company

Initialize from a template

bash

capx init my-company --template=saas-agency
# ✓ Created my-company/company.yaml
# ✓ Created my-company/playbooks/ (12 playbooks)
# ✓ Created my-company/.capx/ (local state)

This creates a directory with your company configuration:

Directory structure

my-company/
├── company.yaml          # Company configuration
├── playbooks/            # Workflow definitions
│   ├── content-pipeline.yaml
│   ├── seo-audit.yaml
│   ├── client-onboarding.yaml
│   └── ... (12 total)
└── .capx/                # Local state (git-ignored)

Configure your company

Open company.yaml and review the configuration. The template provides sensible defaults, but you can customize everything:

company.yaml

yaml

name: my-company
description: AI-powered marketing agency

agents:
  - role: strategist
    adapter: claude
    model: claude-sonnet-4-6
    budget: 200
    heartbeat: "0 9 * * *"        # Daily at 9 AM

  - role: engineer
    adapter: claude
    model: claude-sonnet-4-6
    budget: 300
    heartbeat: "*/30 * * * *"     # Every 30 minutes

  - role: marketer
    adapter: openai
    model: gpt-4o
    budget: 150
    heartbeat: "0 9 * * MON-FRI"  # Weekdays at 9 AM

  - role: support
    adapter: openai
    model: gpt-4o-mini
    budget: 100
    heartbeat: "*/15 * * * *"     # Every 15 minutes

governance:
  approval_required: true
  auto_approve_below: 10          # Credits
  spend_cap: 500                  # Monthly per company
  kill_switch: enabled

playbooks:
  - content-pipeline
  - seo-audit
  - client-onboarding
  - support-triage

Note

Each agent needs a role, adapter, and budget at minimum. The heartbeat schedule determines how often the agent wakes up to check for work. Cron expressions follow standard POSIX format.

Deploy

Deploy your company

bash

capx deploy
# Deploying my-company...
# ✓ Runtime provisioned (isolated container)
# ✓ Database created (dedicated)
# ✓ Agents activated (4/4 online)
# ✓ Governance: approval queue enabled
# ✓ Spend cap: $500/month
# ✓ Playbooks loaded: 4
# ✓ Live at my-company.capx.ai
#
# Your company is running. Agents are working.

Check status

View company status

bash

capx status
# COMPANY      AGENTS  STATUS    COST/TODAY  TASKS
# my-company   4/4     running   $0.00       0 completed, 0 pending

capx agents list
# ROLE        ADAPTER  MODEL              BUDGET  USED  STATUS
# strategist  claude   claude-sonnet-4-6  $200    $0    active
# engineer    claude   claude-sonnet-4-6  $300    $0    active
# marketer    openai   gpt-4o             $150    $0    active
# support     openai   gpt-4o-mini        $100    $0    active

Trigger your first playbook

Run a playbook manually

bash

capx playbooks run content-pipeline --input topic="AI infrastructure"
# Running content-pipeline...
# Step 1/3: research (strategist) ✓ 12s, $0.42
# Step 2/3: draft (marketer) ✓ 8s, $0.38
# Step 3/3: review (strategist) → awaiting approval
#
# Total cost: $0.80
# Approval required: review the output at my-company.capx.ai

Tip

The playbook paused at the review step because governance requires approval. Check your dashboard or run capx activity to review and approve.