Free API key agent: run an AI agent without pasting keys into code

The problem with keys in code

Most agent setups ask you to paste an API key into an environment file or hard-code it. From there it gets copied into dotfiles, logs, crash reports, and screenshots. One misplaced .env and your key is effectively public.

A free API key agent flips that around. The key lives in a vault you control, and the agent only ever talks to a proxy that holds the key. Your code never contains the secret, so there is nothing to leak.

How cloud key storage works

1. You add a key to the encrypted vault from the dashboard.
2. The agent logs in with a short-lived device token, not the key.
3. When the agent calls a model, it asks the platform to make the request on its behalf.
4. The platform decrypts the key in memory, calls the provider, and returns only the answer.
5. The key itself never reaches your device, your prompts, or your logs.

What it actually costs

The agent and the control plane are free and open source under the MIT license. You pay only for the model tokens you use, at the provider’s normal rate. There is no per-agent fee and no markup on tokens.

The free tier covers a single device and a normal amount of day-to-day use. If you run a fleet, the paid plan adds more devices, higher task limits, and longer history — the vault stays the same.

Setting it up

1. Install the agent: curl -fsSL https://remoteagent.online/install.sh | bash
2. Log in: remoteagent login and follow the link.
3. Open the dashboard, go to Settings → API keys, and add your OpenAI or Anthropic key.
4. Start the agent: remoteagent start, then remoteagent chat "hello".

Why it is safer than a .env

• The device holds a revocable token, not the key.
• Rotating a key is one click, not a find-and-replace across machines.
• Even a fully compromised agent has no key material to exfiltrate.
• Every model call is logged, so surprise spend shows up immediately.

Common questions