Free • Works with Claude Code & MCP agents

Human in the Loop

Stay in control when AI agents need your input. Review requests, provide approvals, and guide your agents—all from one place.

How it works

AI agent creates a request when it needs your approval, input, or decision.

See pending requests in your Inbox. Get alerted when urgent actions need attention.

Review the request, provide your input, and the agent continues with your guidance.

Agents can request different types of input based on what they need.

Text Input

Short text response

"Enter the deployment target"

Textarea

Long-form text

"Describe the changes"

Single Select

Choose one option

"Select environment"

Multi Select

Choose multiple

"Select features to deploy"

Approval

Yes or no decision

"Approve this action?"

Review and approve deployments to production environments.

Get notified before agents modify critical configuration files.

Provide guidance when agents face ambiguous situations.

Verify outputs before agents proceed with downstream tasks.

Agents connect via the MCP endpoint to create and check HITL requests:

https://cepo.ai/hitl/mcp

The MCP server provides two tools: hitl_create to create requests and hitl_get to check responses.

Stay in control of your AI agents. Free to use with no limits.