Lock down your agents.
Protect your system.
We put every AI agent inside a kernel-level vault.
Even if the agent tries to break out, the kernel says no.
Super simple to use. Extremely hard to escape.
Your SSH keys, AWS credentials and .env files stay protected.
And we also tame agents that spiral out of control, burn your tokens, or go rogue.
See it in action
A short tour across install, security, and token monitoring — auto-advances.
By the numbers
- Supported agents
- 5+ CLI agents wrapped: Claude Code, Aider, Gemini CLI, Cursor, Cline
- Detection layers
- 3 Detection layers — pre-exec hooks, PTY parser, filesystem watcher
- Encryption
- 256-bit AES-GCM end-to-end. The relay never sees plaintext.
- Secrets leaving the machine
- 0 Bytes of redacted secrets ever leave your machine
- Sandbox primitives
- 5 Kernel primitives: Landlock, seccomp-BPF, bubblewrap, Job Object, AppContainer
- Languages
- 4 Localized sites — English, Français, Deutsch, Español
- License
- BSL 1.1 Apache 2.0 after 4 years
- Cost
- $0 Self-hosted, no per-seat fee
"The strongest sandbox is the one the agent cannot see. Hooks can be argued with; a kernel cannot."
Build your dream pipeline
Tamer's Master Agent orchestrates multiple workers in parallel. If Gemini finds a bug, it automatically tells Claude to fix it. You just supervise the flow.
Who is tamer for?
Whether you're a solo developer or a team lead, tamer fits your workflow.
-
👨💻
Solo Developers
You run Claude Code or Aider on your laptop and leave it working while you step away. Tamer lets you approve commands and monitor progress from your phone — no need to stay glued to the terminal.
-
👥
Team Leads
Your team runs multiple AI agents across different repos. Tamer's Master Agent orchestrates them, detects conflicts, and ensures nobody's agent wrecks someone else's code. One dashboard, full visibility.
-
🔒
Security-Conscious Orgs
You need AI productivity but can't risk data leaks or supply chain attacks. Tamer's kernel sandbox, E2E encryption, and self-hosted architecture keep your secrets local and your agents contained.
Why tamer.ai?
Master Agent
Supervise multiple agents simultaneously. Detect conflicts, sync workspaces, and manage approvals from one central dashboard.
Remote Human-in-the-Loop (HITL)
Human-in-the-Loop AI for coding agents: approve or reject every bash command, file write, and code change from your phone — even while away from your desk. Learn how HITL works →
Air-Gapped Privacy
Secrets stay local. Sensitive data is redacted before leaving your machine. Your code remains yours.
Human-in-the-Loop (HITL) for AI coding agents
Approve every dangerous bash, file write and code change from your phone — kernel-enforced, end-to-end encrypted.
Security first
Your code, your keys, your data — they never leave your machine unencrypted.
E2E Encryption
AES-256-GCM. Server is a blind relay. Optional passphrase.
3-Layer Protection
Pre-execution hooks, PTY parsing, filesystem watcher.
Secret Redaction
API keys, tokens, .env values never leave your machine.
Self-Hosted
Deploy on your own infra. Zero data leaves your network.
Your safety net
Guardrails that protect your codebase, your tokens, and your sanity.
Git Checkpoint
Automatic snapshots before dangerous commands. One-tap rollback from your phone.
Loop Detector
Detects runaway agents repeating the same failing command. Auto-alert before token burn.
Cost Monitor
Real-time token tracking per session. Configurable cost alerts before your bill explodes.
How it works
Install tamer
curl -fsSL https://server.tamer-ai.dev/install.sh | bash
Connect & run
tamer init && tamer claude
Connect to your server, then wrap any agent — your terminal works exactly as before.
Monitor from anywhere
Open the PWA on your phone — approve, reject, or send input from anywhere.
Architecture
Get started
Three commands to full remote control.