workflow-run

Coordinate multi-stage AI workflows with user approval gates and artifact-based state.

2|Updated Mar 27, 2026
One-click install
npx skills add https://github.com/bdemarzo/ai-workflows --skill workflow-run
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: workflow-run
Source: https://github.com/bdemarzo/ai-workflows/tree/main/skills/workflow-run
Command: npx skills add https://github.com/bdemarzo/ai-workflows --skill workflow-run

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) and assets (resource) components.

What problem does it solve?

It coordinates a multi-stage AI workflow—from ideation to final review and documentation—while keeping the active session as the orchestrator and enforcing user approval gates at key decision points.

Core Features & Use Cases

  • End-to-end workflow orchestration: Delegates to the right stage skills (idea/spec/plan/implement/review) and consolidates outputs.
  • Explicit user gates: Pauses for approvals after idea, spec, plan, implementation-review, and final gap resolution plus docs close-out.
  • Stable artifact-based state: Treats docs/workflows/{slug}/run.md as the source of truth so another session can resume without chat history.

Quick Start

Run the workflow-run skill to guide your AI session through idea creation, structured specification, planning, implementation, review, and documentation close-out with required user approvals between phases.

Frequently Asked Questions about workflow-run

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

FAQPage Schema
How do I orchestrate multi-stage AI workflows with user approval gates between phases?

Multi-stage workflow orchestration with user approval gates delegates tasks across ideation, specification, planning, implementation, review, and documentation phases. The orchestrator pauses execution at each decision point, requiring explicit user approval before proceeding to the next stage.

What is artifact-based state management for durable workflow planning sessions?

Artifact-based state management treats a run.md file as the source of truth for workflow progress. It records durable artifacts and review-round files across stages, allowing another session to resume the workflow from the current state without relying on chat history.

Can I resume a guided workflow implementation in a new session if the chat history is lost?

Yes, you can resume a guided workflow in a new session because the state is persisted as a file ledger. The orchestrator reads the maintained run.md file to reconstruct progress, fixed dossier artifacts, and persona-bound operators without needing the previous chat context.

What's the best way to enforce structured reviews during product planning and implementation?

The best way to enforce structured reviews is using an orchestrator-led delegation model that spawns persona-bound reviewers per phase. It creates dedicated review-round files for each stage, ensuring implementation review and gap resolution are verified through explicit user gates.

Does workflow-run require specific dependencies to manage engineering planning tasks?

No specific dependencies are required to manage engineering planning tasks. The skill operates autonomously to coordinate product and implementation planning, relying only on its internal references and assets to generate durable artifacts and guide stage reviews.

When should I not use a guided orchestration approach for product specification?

You should not use guided orchestration for product specification when a task requires only a single-stage output or lacks durable artifact requirements. The explicit user gates and multi-phase delegation model are designed for complex workflows spanning ideation through close-out documentation.