clarify-confirm-continue

name	clarify-confirm-continue
description	Intake multi-step tasks by clarifying ambiguities, summarizing understanding, and confirming readiness via ask_question before executing code changes.
key_features	["Task intake and scope verification","Shorthand ccc and pre-flight triggers","Compositional grilling for ambiguity resolution","Fast-tracking for VERY obvious assumptions","Structured understanding summary","On-demand artifact generation for complex iteration","Single-click readiness confirmation gate"]

Clarify-Confirm-Continue (ccc / Pre-Flight)

This skill defines a disciplined task intake and pre-flight check workflow for multi-step or complex user requests. It ensures the agent and user are fully aligned before any codebase modification or extensive execution begins.

📦 Prerequisites & Skill Dependencies

• REQUIRED SKILL: grilling MUST be installed alongside clarify-confirm-continue for Tier 2 ambiguity resolution.

When to use this skill

Activate when the user assigns a task and requests a pre-flight check, or uses trigger phrases like:

• "ccc" / "clarify-confirm-continue"
• "pre-flight" / "check with me first" / "before you start"
• "make sure you understand" / "confirm before proceeding"

Procedural Workflow

Phase 1: Clarification & Fact-Finding

1. Investigate Facts First: Use read-only tools (grep_search, view_file, list_dir) to check existing code patterns, architecture, or definitions. Never ask the user for verifiable facts that the codebase can answer directly.
2. Assess Ambiguity (Two-Tier Check):
   • Tier 1 (VERY Obvious): If the task is straightforward and any assumptions or inferences are VERY obvious (things that do not require interactive debate, but are simply worth mentioning for transparency), proceed directly to Phase 2.
   • Tier 2 (Ambiguous / Decisions Needed): If major architectural decisions, multiple interpretations, or non-obvious trade-offs exist, apply the Q&A pattern from grilling:
     • Ask clarifying questions one at a time.
     • Provide your recommended answer for each question.
     • Wait for user feedback on each question before continuing. (Do not present the final summary or readiness confirmation until all questions are answered).

Phase 2: Structured Summary

Once all ambiguities are resolved (or immediately if Tier 1 applies), output a structured summary in your chat response:

• 🎯 Goals: What will be accomplished.
• 🛡️ Non-Goals / Scope Boundaries: What will explicitly be left untouched.
• 📌 Assumptions: Verifiable facts or VERY obvious inferences worth mentioning.
• 🛠️ Execution Plan: High-level steps to be performed once approved.

Phase 3: Readiness Gate (ask_question)

Immediately following the summary in the same turn, invoke the ask_question tool with a single, mandatory confirmation check:

• Question: "I have summarized my understanding above. How would you like to proceed?"
• Options:
  • "(Recommended) Yes, proceed as summarized."
  • "Open this proposal in a markdown artifact so we can iterate on specific sections."
  • "No, let's adjust the scope or clarify further in chat."

Phase 4: Handling Gate Selection

• If "Yes, proceed": Transition immediately to task execution.
• If "Open in artifact": Use write_to_file to save the structured summary and execution plan as a markdown artifact (.md file) in the session's artifact directory. Stop calling tools and wait for the user's line-by-line review and comments.
• If "No, adjust in chat": Ask what needs adjustment or loop back to Phase 1.

Critical Rule: Zero Execution Before Approval & Global Rules

• NO CODE MODIFICATIONS: Do not edit project files, create branches, or run mutating commands during Phase 1, 2, or 3.
• STRICT GATING: You must receive explicit approval via ask_question (or clear write-in approval) before transitioning to task execution.
• GLOBAL RULES UNCHANGED: Approval at the Phase 3 gate simply confirms that you understand the task and authorizes you to begin researching or drafting code. Nothing changes about global Git/GitHub write restrictions or commit rules in AGENTS.md—all standard outward-facing write approvals remain strictly and independently enforced during execution.