BE1st Command Center

Command Center

Manage your site with AI-powered terminal. No extra costs with Claude Max.

Select Site

or start from scratch

Terminal

Hover presets for details · Click to run

QUICK:

Auto-Pilot Idle

Autonomy Level 3 - Guided

Observe Suggest Guided Full Auto

Applies fixes one at a time with 3s pause between each. You can stop anytime.

Budget Cap $5.00

Claude API Key Required for Auto-Pilot

Auto-Pilot Ready

Set your autonomy level and budget, then hit Start.

Command Center Guide

This is your site's control room. Preview pages, edit with AI, and deploy changes — all from one screen.

How Editing Works

Your site ships with a full Claude AI context file (CLAUDE.md) that teaches the AI everything about your site — pages, design system, file structure, and how to make changes correctly.

There are two ways to edit:

Terminal (below the preview) — Type claude to start Claude AI. Describe what you want in plain English. Uses your Claude Max subscription ($0 per edit).
Preset Buttons — Click + Page, Colors, Theme, Edit, Feature, or Content above the terminal to run slash commands that give Claude structured instructions.

Pro Tip: MCP Tools

You can supercharge Claude with MCP servers — plugins that give Claude access to external tools. Two powerful ones for site editing:

21st.dev (Magic MCP) — Gives Claude access to a library of 1,000+ premium UI components. When you say "add a pricing section", Claude can pull in a beautiful pre-built component instead of writing from scratch.
Context7 — Gives Claude access to up-to-date library documentation (React, Tailwind, etc.) so it always uses the latest APIs.

Every BE1st site ships with both MCPs pre-configured in .claude/settings.local.json. Context7 works immediately. For 21st.dev, get your API key at 21st.dev/magic/console ($20/mo after 5 free uses) and replace the placeholder in the settings file.

Auto-Pilot Mode

Auto-Pilot uses AI to automatically audit your site for issues (broken layouts, missing content, accessibility gaps, design inconsistencies) and fix them — hands-free.

Requires your own Claude API key (not Max subscription) since it makes multiple API calls in a loop. Enter your key in the Auto-Pilot panel before starting.

Observe Read-Only

Audits your entire site and reports every issue it finds — severity, file, description, and suggested fix. Makes zero changes. Perfect for understanding what needs work before committing to fixes. One Claude call (~$0.01-0.03).

Suggest Diffs Only

Audits AND generates the actual code fixes for each issue, but does not apply them. You see exactly what would change before anything touches your files. Review each suggestion in the log. Multiple Claude calls (audit + one per issue).

Guided Recommended

Audits, generates fixes, and applies them one at a time with a 3-second pause between each. You can watch each fix land in the preview and hit Stop anytime. After all fixes, rebuilds the frontend. One pass — does not re-audit.

Full Auto Advanced

The full loop: audit → fix all issues → rebuild → re-audit → repeat. Keeps going until the site is clean (zero issues) or hits your budget cap. Up to 10 iterations. Set a budget cap — typical full cleanup costs $0.50-2.00.

Cost Estimate

Each audit call: ~$0.01-0.03. Each fix: ~$0.01-0.05. A typical Level 4 run with 8-12 issues across 2-3 iterations: ~$0.50-1.50. The budget cap ensures you never spend more than you set.

Terminal Commands

claude                          # Start interactive Claude AI session
claude "Make the hero darker"   # One-shot edit command
claude /add-page                # Run a preset slash command
npx vite build                  # Rebuild the frontend after changes

Keyboard Shortcuts

The terminal supports standard terminal shortcuts. Type exit or press Ctrl+C to stop a running Claude session.

How Costs Work

Terminal / Claude CLI — Uses your Claude Max subscription. $0 per edit.
Auto-Pilot — Uses your Claude API key (BYOK). Billed per-token by Anthropic. Budget cap protects you.
Preset buttons — Run through CLI. $0 per edit.