What is LangGraph HITL?

A mechanism to pause workflows and require human input at defined points using LangGraph HITL patterns.

What happens on timeout?

If no approval is received within the configured timeout, the escalation path is triggered or the workflow pauses until a higher authority responds.

Does this require persistence?

Yes, persistence via langgraph-checkpoint ensures state is saved during interruptions.

langgraph-hitl

npx machina-cli add skill a5c-ai/babysitter/langgraph-hitl --openclaw

Files (1)

SKILL.md

1.2 KB

LangGraph Human-in-the-Loop Skill

Capabilities

Implement interrupt points in LangGraph workflows
Configure human approval gates
Design intervention interfaces
Handle workflow resumption after human input
Implement timeout and escalation logic
Create notification systems for pending approvals

Target Processes

langgraph-workflow-design
tool-safety-validation

Implementation Details

HITL Patterns

Interrupt Before: Pause before node execution
Interrupt After: Pause after node for review
Conditional Interrupt: Context-based pausing
Tool Call Approval: Review tool invocations

Configuration Options

Interrupt node selection
Approval timeout settings
Escalation paths
Notification channels
State modification permissions

Best Practices

Clear approval prompts
Reasonable timeout defaults
Proper escalation chains
Audit logging for approvals

Dependencies

langgraph
langgraph-checkpoint (for persistence)

Source

git clone https://github.com/a5c-ai/babysitter/blob/main/plugins/babysitter/skills/babysit/process/specializations/ai-agents-conversational/skills/langgraph-hitl/SKILL.md

View on GitHub

Overview

This skill adds interrupt points, approval gates, and intervention interfaces to LangGraph workflows. It enables pausing before or after nodes, handles timeout and escalation, and resumes workflow execution after human input. It also supports notifications and audit logging for approvals.

How This Skill Works

HITL patterns are implemented as Interrupt Before, Interrupt After, Conditional Interrupt, and Tool Call Approval. Configuration options include interrupt node selection, approval timeout, escalation paths, notification channels, and state modification permissions. The system persists workflow state and resumes automatically once human input is provided or escalation criteria are met.

When to Use It

When a critical node could cause unsafe outcomes and requires human sign-off before execution.
When a node finishes and needs post-action review before proceeding.
When a tool invocation is sensitive and requires explicit approval.
During design or validation phases of a LangGraph workflow to ensure safety checks.
When approvals timeout or escalate to next authority to avoid stalls.

Quick Start

Step 1: Identify critical nodes and insert HITL Interrupt points in the LangGraph workflow.
Step 2: Configure approval timeouts, escalation paths, and notification channels.
Step 3: Test the end-to-end flow by simulating human input and resumption.

Best Practices

Craft clear, actionable approval prompts.
Set reasonable timeout defaults and test them.
Define explicit escalation paths and notification channels.
Enable audit logging for all approvals and interventions.
Test resume flows end-to-end to prevent deadlocks.

Example Use Cases

Pause a workflow before deleting or modifying critical data until a human approves.
Require manager sign-off for high-risk tool calls.
Pause after a data transformation for quality validation.
Escalate unapproved actions to a second reviewer after timeout.
Provide an intervention UI to adjust parameters mid-execution.

Frequently Asked Questions

Add this skill to your agents