clarification-broker

Compress planner candidate questions into a batched frontdesk clarification set and normalize answers.

3.5k|342|Updated Oct 25, 2025
One-click install
npx skills add https://github.com/bfly123/claude_code_bridge --skill clarification-broker
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: clarification-broker
Source: https://github.com/bfly123/claude_code_bridge/tree/main/docs/plantree/plans/agentic-loop-workflow/drafts/agentroles.ccb_clarification_broker/skills/clarification-broker
Command: npx skills add https://github.com/bfly123/claude_code_bridge --skill clarification-broker

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

In multi-agent planning workflows, a planner can generate many overlapping or premature questions that overwhelm the user-facing frontdesk. This Skill consolidates those candidate questions into one staged, deduplicated clarification batch and normalizes the user's answers back into a form the planner can consume.

Core Features & Use Cases

  • Question Batching: Merges candidate questions into a single stage-batched set instead of interrupting the user repeatedly.
  • Deduplication and Deferral: Removes duplicates and already-answerable questions, and defers later-phase questions to avoid overloading the frontdesk.
  • Answer Normalization: Converts user responses, applied defaults, and deferred items back into a normalized structure for the planner.
  • Use Case: During an agentic planning loop, the planner produces twelve raw questions; this Skill reduces them to four essential frontdesk questions, applies known defaults, and returns normalized answers once the user replies.

Quick Start

Use the clarification-broker skill to turn these planner candidate questions into one batched frontdesk question set and normalize the user's answers.

Frequently Asked Questions about clarification-broker

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

FAQPage Schema
How do I batch clarification questions in a multi-agent planning workflow?

Collect the planner's candidate questions, remove duplicates and already-answerable items, then present them as one stage-batched set to the frontdesk. This Skill performs that compression and returns normalized answers after the user responds.

How to reduce user interruptions from an AI planner agent?

Prefer a single stage-batched question set over many interruptions. Defer questions that belong to later phases and apply known defaults so the frontdesk only asks what is essential for the current stage.

What inputs does a clarification broker need?

It needs the candidate questions, a task packet draft, existing decisions and user constraints, and any previous answers. With these it produces a user-facing question batch, applied defaults, deferred questions, and normalized answers.

Can the clarification broker modify the task state directly?

No. The Skill explicitly does not alter authoritative task state directly. It only produces the question batch and normalized answers, leaving state changes to the planner or orchestrating workflow.

When should questions be deferred instead of asked?

Defer questions that belong to later phases of the plan or that would overload the frontdesk in the current stage. This keeps each clarification round focused and avoids premature decisions.