ask-me

Coordinate human clarification via AUQ for blocked agentic workflow decisions.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/cometjc/dev-skills --skill ask-me-cometjc
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ask-me
Source: https://github.com/cometjc/dev-skills/tree/main/skills/ask-me
Command: npx skills add https://github.com/cometjc/dev-skills --skill ask-me-cometjc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill resolves ambiguous, risky, or execution-blocked decisions by coordinating structured user clarification through AUQ, preventing the workflow from guessing when user intent is required.

Core Features & Use Cases

  • AUQ orchestration contract: Ensures correct session initialization, session-level project/context propagation, and reliable answer-based resume semantics.
  • Blocking vs non-blocking routing: Uses blocking AUQ only when a slice is truly gated, while allowing independent slices to continue.
  • Timeout-safe recovery: Avoids re-asking the same question after a timeout and instead recovers the quiz/session identifiers to fetch the still-pending answer.
  • Cursor-friendly Await fallback: When tool waits time out, uses a shell-based blocking fetch to let Cursor agents recover answers without re-entering the same timed-out path.
  • Prompt discipline & governance rules: Enforces AUQ as the clarification mechanism (not plain text menus) and applies path-first question ordering for rule/content governance updates.

Quick Start

Ask the user to resolve an ambiguous routing or high-risk decision when execution is blocked by missing intent by using the ask-me skill in your workflow.

Frequently Asked Questions about ask-me

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

FAQPage Schema
How do I handle human-in-the-loop clarification in agentic workflows without blocking independent tasks?

Human-in-the-loop clarification can be handled by issuing blocking ask_user_questions only for truly gated slices, allowing independent workflow slices to continue executing safely without derailing the overall flow.

What is the best way to recover an agentic workflow after an MCP or AUQ timeout?

Timeout recovery in agentic workflows requires fetching the still-pending answer using recovered quiz or session identifiers, avoiding re-asking the same blocking question and resuming execution from the recovered answered state.

How does human-in-the-loop clarification work with Cursor integration when tool waits time out?

Cursor integration uses a shell-based blocking fetch fallback when tool waits time out, enabling Cursor agents to recover pending answers without re-entering the same timed-out execution path.

Why does my workflow re-ask the same blocking question after a session timeout?

Workflows re-ask blocking questions after timeouts when AUQ session initialization fails; enforcing correct get_session_id usage ensures session-level project and context propagation for reliable answer-based resume semantics.

Do I need structured AUQ for routing decisions or can I use plain text menus for clarification?

Structured AUQ is required for clarification in question-gated orchestration flows, as plain text menus lack the deterministic blocking behavior and path-first question ordering needed for safe workflow governance.

When should I use blocking questions in workflow orchestration?

Use blocking questions in workflow orchestration when a slice is truly gated by ambiguous routing, risky trade-offs, or missing user intent, ensuring execution is safely blocked until structured answers are provided.