Agents in IT ops. Without the 3am incident.
Give agents deploys, incidents and infrastructure. The change window, the secrets and the blast radius stay yours. TapPass holds what needs a person and records what happened.
✓No credit card needed
Ship fast. Break nothing.
One eager agent can deploy outside the window, rotate the wrong secret or delete a resource. Not out of malice. Because nothing stopped it.
Claude Code, governed. Nothing to install.
Add the TapPass hooks to your Claude Code settings. Every prompt and every tool call is checked before it runs, and why the agent stopped lands on the record.
{
"hooks": {
"UserPromptSubmit": [ … TapPass check … ],
"PreToolUse": [ … TapPass check … ],
"PostToolUse": [ … record only … ],
"Stop": [ … record only … ]
}
}Can TapPass hold a production deploy outside the change window?
Yes. A rule on deploys outside your window holds the action for the person on call, who approves or denies it with a passkey on their own device.
Do secrets end up in the record?
No. Keys and tokens are redacted before a model sees them, and the record keeps redacted content and metadata only.
Does it work with coding agents like Claude Code?
Yes. Claude Code is governed through its hooks: every prompt and tool call is checked, and why the agent stopped is recorded. Other clients connect through MCP.
Can we send the record to our SIEM?
Yes. Every decision streams out as OpenTelemetry to Splunk, Microsoft Sentinel or Elastic.
Give your agents the pager. Keep the change window.
Connect one agent and see every call TapPass would have held, stopped or routed elsewhere. Nothing is blocked while you learn.