coordinator-mode

Orchestrates parallel multi-agent workflows with decomposition, dispatch, and synthesis protocols.

2|Updated May 30, 2026
One-click install
npx skills add https://github.com/virahitvin8/crafty-gis --skill coordinator-mode-virahitvin8
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: coordinator-mode
Source: https://github.com/virahitvin8/crafty-gis/tree/main/GIT_STAR/.agent/skills/coordinator-mode
Command: npx skills add https://github.com/virahitvin8/crafty-gis --skill coordinator-mode-virahitvin8

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Complex tasks often require multiple agents working across different domains, but naive sequential execution wastes time and poorly briefed workers produce wrong results. This Skill provides a structured coordination mode that decomposes tasks, dispatches parallel workers safely, and synthesizes their outputs into coherent results. ## Core Features & Use Cases - Phase-Based Orchestration: Enforces a Research → Synthesis → Implementation → Verification lifecycle with explicit concurrency rules for each phase. - Worker Prompt Templates: Provides research, implementation, and verification prompt patterns that prevent the "never delegate understanding" anti-pattern. - Fork and Spawn Semantics: Defines when to fork context-inheriting agents versus spawning fresh specialists, plus continue-vs-respawn decision rules. - Use Case: When asked to refactor an authentication system, decompose it into parallel codebase research workers, synthesize findings into a plan, dispatch sequential implementation workers per file set, then run independent verification workers. ## Quick Start Activate coordinator mode to break my complex task into parallel worker subtasks and synthesize their results into a unified plan.

Frequently Asked Questions about coordinator-mode

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

FAQPage Schema
How do I coordinate multiple AI agents on one task?

Decompose the task into worker subtasks, classify each as research, implementation, or verification, then dispatch read-only workers in parallel and write workers sequentially. Synthesize all worker outputs yourself before reporting results.

How to write effective prompts for worker agents?

Include specific file paths, line numbers, and exact changes rather than vague instructions like "fix the bug." Brief the worker with context about the goal, what you already checked, scope boundaries, and the expected output format.

When should agents run in parallel versus sequentially?

Run agents in parallel when they read different files or perform independent checks like audits and tests. Run them sequentially when two agents write the same file or when one implementation depends on another agent's output.

What is the difference between forking and spawning an agent?

Forking inherits your full context, suiting open-ended research with short directive prompts. Spawning creates a fresh specialist with clean context, better for specialized domain work or independent second opinions.

Why do multi-agent workflows produce wrong results?

Failures usually come from vague worker prompts that delegate understanding, skipping the research phase, or fabricating results before workers finish. Launch only 2-5 workers per round and wait for actual completion notifications.