create-workflow-command

Creates workflow commands that orchestrate multi-step execution through sub-agents with file-based task prompts.

1.5k|154|Updated Nov 13, 2025
One-click install
npx skills add https://github.com/NeoLabHQ/context-engineering-kit --skill create-workflow-command
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: create-workflow-command
Source: https://github.com/NeoLabHQ/context-engineering-kit/tree/main/plugins/customaize-agent/skills/create-workflow-command
Command: npx skills add https://github.com/NeoLabHQ/context-engineering-kit --skill create-workflow-command

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Embedding detailed step instructions directly in a command file bloats the orchestrator's context window and degrades agent performance. This Skill solves the context bloat problem by generating workflow commands that keep the orchestrator lean while sub-agents load detailed instructions from separate task files on demand.

Core Features & Use Cases

  • Context-Isolated Orchestration: Generates a lean orchestrator command (~50-100 tokens per step) that dispatches sub-agents, while detailed instructions (~500+ tokens) live in separate task files.
  • Multiple Execution Patterns: Supports sequential steps, parallel independent steps, and stateful multi-step workflows using the resume parameter.
  • Complete Scaffolding: Creates the tasks/ directory structure, per-step task files with goals and success criteria, the orchestrator command with frontmatter, and optional custom agents.
  • Use Case: You want a feature-implementation workflow with research, architecture, and coding phases. This Skill generates the orchestrator command plus three self-contained task files that sub-agents read and execute in isolated contexts.

Quick Start

Ask the agent to create a workflow command named feature-implementation that runs research, architecture, and implementation steps through sub-agents.

Frequently Asked Questions about create-workflow-command

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

FAQPage Schema
How do I create a multi-step workflow command with sub-agents?

Define the workflow name, description, and steps, then create a lean orchestrator command that dispatches sub-agents via the Task tool. Each step's detailed instructions go in a separate task file under tasks/ that the sub-agent reads with the Read tool.

How do I avoid context bloat in agent workflow commands?

Store detailed step instructions in separate task files instead of embedding them in the command. The orchestrator stays at roughly 50-100 tokens per step while sub-agents load the full 500+ token instructions from files on demand.

Can sub-agents spawn their own sub-agents or load skills?

No. Sub-agents cannot nest the Task tool and skills do not auto-load inside sub-agents. Keep all orchestration in the main command and pass explicit file paths or inline context to sub-agents.

How do I run workflow steps in parallel with sub-agents?

Launch multiple agents simultaneously in one phase, each reading its own task file, then wait for all results before a synthesis step. This pattern suits independent analyses like security, performance, and code quality running concurrently.

How do I share context between steps in a sub-agent workflow?

Either pass summaries of previous step outputs in the next dispatch prompt, or capture the agent_id from a step and use the resume parameter to continue the same agent context with additional instructions.