pipeline

Execute predefined multi-agent pipelines with gates and step handoffs.

Updated Jun 1, 2025
One-click install
npx skills add https://github.com/Yokhan/yokhanFitnessDocsWebsite --skill pipeline-yokhan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pipeline
Source: https://github.com/Yokhan/yokhanFitnessDocsWebsite/tree/main/.claude/skills/pipeline
Command: npx skills add https://github.com/Yokhan/yokhanFitnessDocsWebsite --skill pipeline-yokhan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coordinating and executing multi-agent workflows requires managing step sequencing, handoffs, and gating between agents. This skill provides a structured runner that reads predefined pipelines and orchestrates agent steps with clear handoffs and observable outcomes.

Core Features & Use Cases

  • Pipeline orchestration: Define pipelines as ordered steps with gates that control progression.
  • Handoff and persistence: Each step writes a summary to tasks/current.md for downstream agents to consume and audit.
  • Gate-driven control: Supports user_approval, typecheck, tests_pass, and verdict gates to ensure quality before proceeding.
  • Use Case: Automate a feature development lifecycle from research to implementation to review using a single pipeline file.

Quick Start

Provide a pipeline name to run and the runner will execute steps in order.

Frequently Asked Questions about pipeline

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

FAQPage Schema
How do I coordinate a multi-agent workflow with sequential handoffs?

Multi-agent workflow coordination uses predefined pipelines stored in .claude/pipelines to execute ordered steps. The runner orchestrates agent steps with structured handoffs and writes summaries to tasks/current.md.

What is a gate in a multi-agent pipeline and when do I need one?

A gate in a multi-agent pipeline controls progression between steps by enforcing quality checks. You need gates like user_approval, typecheck, tests_pass, or verdict to ensure conditions are met before the next agent executes.

How do I automate a feature development lifecycle from research to review?

Automate a feature development lifecycle by creating a single pipeline file that chains research, implementation, and review steps. The runner orchestrates these phases with gate-driven control and end-to-end metrics recording.

Can I use custom gating conditions to control step execution in an automated pipeline?

You can control automated pipeline execution using supported gate types. The runner handles user_approval, typecheck, tests_pass, and verdict gates to ensure quality and correct conditions before proceeding.

Does running a predefined pipeline require any external dependencies?

Running predefined pipelines requires no external dependencies. The runner operates independently by reading pipeline definitions from .claude/pipelines and coordinating agent steps internally without external components.

Why are downstream agents in my pipeline missing context from previous steps?

Downstream agents miss context when step summaries are not persisted to tasks/current.md. The pipeline runner writes each step's output to this file so downstream agents can consume and audit the handoff data.