ask-user

Present 2-4 labeled options to users and gate agent execution until they respond.

Updated Aug 17, 2026
One-click install
npx skills add https://github.com/AMC-JTC/gbrain-1 --skill ask-user-amc-jtc
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ask-user
Source: https://github.com/AMC-JTC/gbrain-1/tree/main/plugin/skills/ask-user
Command: npx skills add https://github.com/AMC-JTC/gbrain-1 --skill ask-user-amc-jtc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? LLM agents often guess at ambiguous decisions or proceed with silent defaults, wasting work and eroding trust. This Skill provides a canonical pattern for pausing execution at decision points, presenting clear choices, and waiting for explicit human input before continuing. ## Core Features & Use Cases - Choice Gate Pattern: Present 2-4 self-explanatory options with a mandatory escape hatch (Skip/Cancel), then stop the turn completely until the user responds. - Platform-Agnostic Formatting: Works on Telegram inline buttons, Discord, Slack, CLI, or any agent with a message tool, including OpenClaw's native clarify tool. - Response Handling: Maps button clicks, numbers, fuzzy text matches, and unrelated replies to the correct branch. - Use Case: During a cold-start import, the agent asks whether to import Google Contacts via ClawVisor, direct OAuth, a Takeout export, or skip — then waits for the user's pick before proceeding. ## Quick Start Ask the user how to file these meeting notes, presenting merge, new page, and skip options, then wait for their response before acting.

Frequently Asked Questions about ask-user

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make an AI agent wait for user input before proceeding?

Present 2-4 numbered options with a clear question, then stop the turn entirely with no further tool calls or follow-up messages. The user's response triggers the next turn, where you acknowledge briefly and branch on their choice.

How many options should a choice prompt include?

Include 2-4 options maximum, since more than four causes decision paralysis. Always add an escape hatch like Skip or Cancel as the final option so the user can decline every path.

Does this choice gate pattern work on Telegram and Discord?

Yes, the pattern is platform-agnostic. Use inline buttons on Telegram, Discord, or Slack when supported, plain numbered options in any chat interface, or the native clarify tool on OpenClaw agents.

When should an agent not ask the user before acting?

Skip the gate for clear unambiguous instructions, low-stakes decisions where a wrong guess costs nothing, time-critical operations, and cases where the user already stated a preference. Reserve gates for forks where rework is expensive.

How do I handle text replies instead of button clicks?

Map responses flexibly: 'the first one', 'A', or '1' map to the first option, partial words fuzzy-match against option labels, and 'none of those' triggers alternatives. An unrelated message means the user moved on, so drop the gate.