Can I integrate AgentPing into automated scripts?

Yes, AgentPing is designed for automation, offering a --json flag and specific exit codes (0 for approval, 1 for denial, 2 for timeout) that make it easy for scripts and AI agents to parse responses.

How does AgentPing improve AI safety in Claude Code?

It provides a standardized 'human-in-the-loop' mechanism, ensuring agents pause for explicit verification before performing high-risk actions like deleting files or migrating databases.

Is the AgentPing Web UI accessible?

Absolutely. The Web UI is fully ADA compliant, meeting WCAG 2.1 AA standards for color contrast, keyboard navigation, and screen reader support.

Does AgentPing support multi-step approvals?

Yes, the 'approve-steps' command allows agents to present a complex plan with multiple items, which the human can approve in bulk or partially validate.

What is AgentPing used for?

AgentPing is a protocol and toolset that allows AI agents to send structured requests—like yes/no approvals or multiple-choice questions—to humans across various platforms including CLI and Web UIs.

AgentPing

Name: AgentPing
Author: lev-os

bylev-os

0•

Collaboration Tools

Facilitates structured, multi-channel communication between AI agents and humans for approvals, decision-making, and feedback loops.

AgentPing bridges the gap between autonomous AI agents and human oversight by providing a robust, structured interaction protocol. It enables agents to move beyond unstructured text prompts by generating typed 'pings' for approvals, multiple-choice selections, and multi-step checklists that render consistently across CLI, Web, and messaging environments. This skill is essential for maintaining safety and accuracy in complex workflows, allowing Claude to pause for human direction, verify destructive actions, or gather specific preferences using a production-ready, JSON-compatible interface with built-in daemon management.

Key Features

01Comprehensive daemon management for handling asynchronous feedback and long-running agent tasks.

02Fully ADA-compliant (WCAG 2.1 AA) 'Cyber-Premium' dark theme Web UI and Storybook documentation.

03Multi-surface rendering support for CLI, Web UI (React/Tailwind), Slack, and Telegram.

04Standardized interaction types including Approval, Select, Ask, and multi-step Step-Approval.

05Agent-first JSON interface with predictable exit codes for automated workflow integration.

060 GitHub stars

Use Cases

01Gathering specific research directions or priority constraints from a user during complex, multi-stage agent operations.

02Implementing safety gates that require human sign-off before executing destructive commands or production deployments.

03Presenting structured architectural or framework choices to a user for selection during automated code generation.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lev-os/agents agentping

For use in Claude.ai and ChatGPT

Download Skill