11 Agents.
One Dashboard.
Your Phone.
Approve risky permissions, review code diffs, and watch token spend in real time. From your phone, end-to-end encrypted.
Pairs with the 11 CLI agents already in your terminal
Three problems the API dashboard does not solve.
Five terminals, zero unified view
Claude Code in one window. Codex in another. Aider in a third. No single place shows what is active, what is stuck, or which session is burning tokens on a loop. You context-switch between terminals to answer 'what is happening right now.'
Bills you find out about on the 1st
A runaway loop overnight. A context window blown out at 3am. An unattended Aider session that recompiled its prompt 400 times. The API invoice arrives weeks later. Nothing on your dashboard could have stopped it.
Approvals stuck behind a closed laptop
Your agent pauses for permission. You are in a meeting, on a flight, walking the dog. The agent waits for hours, or you preemptively grant blanket approval and lose the safety net entirely.
Built for the way developers actually run AI agents.
Remote control, session memory, and full cost attribution for all 11 agents in one encrypted view.
11 agents, one live view
Watch every CLI agent at once. Live status across Claude Code, Codex, Gemini CLI, OpenCode, Aider, Goose, Amp, Crush, Kilo, Kiro, and Droid. Per-session error attribution, token spend, and machine health, all in one encrypted view.
Phone in your pocket, agent on your desk
Approve permissions in three taps. Send voice commands hands-free. Review diffs from a coffee shop. Your agents stop waiting on you.
Pick up any session, on any machine
Checkpoint, resume, share, replay. Every message encrypted with TweetNaCl on your device before it leaves.
Token-level cost, not month-end surprise
Per-message cost across 300+ models. Budget caps that throttle or kill the agent before the damage is done.
See what your fleet did this week
GitHub-style heatmap across every session. Find your heavy days, your idle days, and the agent burning the most cycles.
Notified the moment a session lands
Push when a job completes, errors, or stalls. No more refreshing the terminal from the kitchen.
Send traces where your team already looks
Stream session metrics to Grafana, Datadog, Honeycomb, or New Relic. The same observability stack your platform team is already paying for.
Approve a diff without your laptop open
Syntax-highlighted diffs, inline. Approve, comment, or kick back from your phone.
Walk into Monday with the rundown
Sunday morning, your inbox: 47 commits, the auth refactor landed Thursday, three sessions caught a regression. Coffee in hand, full picture, no scrolling. Hit summarize on any session for the on-demand version.
Approve, dictate, review. From your phone, with full diff context.
Three workflows that disappear when you can do them on a 6-inch screen instead of a 16-inch one.
claude-code wants to write to src/auth/tokens.ts
Permission Approval
Full diff plus risk badge before you tap. Not a blind yes or no.
Listening...
You said
“Stop the codex session”
Executed
codex session terminated
Voice Commands
Stop a session, kick a build, dictate a prompt. Hands stay on the wheel.
Code Review
Inline diff, syntax highlighting, one-tap approve. Same review you would do at your desk.
Your agents work around the clock. You do not have to.
Sessions stay end-to-end encrypted on your device, even when you are reviewing them from a phone.
Cost visibility before the invoice arrives.
Per-agent, per-session, per-message spend across all 11 agents. Updated on every page load, not at the end of the month.
Per-agent, per-session, per-message
Tag sessions by client or project. See exactly which session ate $40 today, not which day of the month it happened. Sort, filter, and group across all 11 agents in one view.
Budget caps that pull the plug
Set daily or monthly limits per agent. Styrby warns at your threshold, throttles the model, or terminates the session automatically. The 2am invoice problem stops being your problem.
300+ models, current pricing
Token-level breakdowns with input, output, and cache costs. Pricing tables update as providers change them. Find the rogue $40 session inside a $200 month, then set a guardrail so it never happens twice.
Three Steps. Ninety Seconds. Done.
Install one CLI alongside the agents you already use. Pair your phone. Start controlling everything from anywhere.
Install the CLI
One npm command alongside your existing setup. No agent reconfiguration. Styrby wraps your CLI tools without touching their config files.
Sign in and pair your phone
GitHub, Google, or email. Then scan a QR code. The CLI walks you through the end-to-end key exchange in under sixty seconds.
Run from your phone
Live session status, real-time costs, permission approvals, and code review diffs. All encrypted on your device. All available wherever you have a phone.
Pro when you ship solo. Growth when your team scales.
Start on Pro for $39/mo. Move to Growth the moment a second developer joins. Cancel anytime.
Questions you should ask before trusting any tool with your agent data
Can you see my code or prompts?
No. Styrby uses end-to-end encryption (TweetNaCl) with a zero-knowledge architecture. Your code and prompts are encrypted on your machine before anything leaves it. Our servers only process metadata: costs, timestamps, and session status. We cannot read your data, even if compelled to. Exported sessions remain encrypted, and shared session links require a separate key you control.
Tap a question above to read the answer here.