workflow-skill-creator

Distill completed workflows into reusable agent skills with design artifacts.

2.7k|283|Updated May 13, 2026
One-click install
npx skills add https://github.com/google-deepmind/science-skills --skill workflow-skill-creator
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: workflow-skill-creator
Source: https://github.com/google-deepmind/science-skills/tree/main/skills/workflow_skill_creator
Command: npx skills add https://github.com/google-deepmind/science-skills --skill workflow-skill-creator

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Distills an already-completed workflow or interaction into a reusable agent skill, so you can repeat the same multi-step scientific process reliably without re-explaining it each time.

Core Features & Use Cases

  • Workflow distillation: Captures purpose, inputs/outputs, step structure, and decision points from an interaction that already happened.
  • Dependency-aware reuse: Prompts you to identify overlapping existing science skills and ensures the new skill references them instead of reimplementing.
  • Implementation-ready plan: Produces a Phase-2 design document that includes directory structure, required scripts (if any), rate-limiting strategy, and error handling per step.

Quick Start

Tell the agent: “Here’s a workflow I just completed for [task]; help me make it into a reusable skill.”

Frequently Asked Questions about workflow-skill-creator

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

FAQPage Schema
How do I turn a completed workflow into a reusable skill?▼

To turn a completed workflow into a reusable skill, you provide the agent with the finished interaction. The skill distills its purpose, scope, inputs, outputs, steps, and constraints into an implementation-ready design document.

How does workflow distillation handle dependencies for scientific research processes?▼

Workflow distillation handles dependencies by prompting you to identify overlapping existing science skills. It ensures the new skill references these existing bundle skills instead of reimplementing them.

Can I create an agent skill with custom API calls and rate limiting?▼

Yes, you can create an agent skill with custom API calls and rate limiting. The skill produces a Phase-2 design document that includes required scripts, directory structure, rate-limiting strategy, and error handling per step.

What is the brainstorming phase before generating an agent skill?▼

The brainstorming phase is a mandatory step where the agent analyzes your completed workflow. It waits for your explicit approval before proceeding to generate the implementation-ready plan and design artifacts.

Do I need to install any dependencies to distill a workflow into a skill?▼

No dependencies are required to distill a workflow into a skill. The process works within the agent environment to extract your process structure and generate a reusable design document.

What is the best way to document error handling for recurring scientific processes?▼

The best way to document error handling for recurring scientific processes is using a Phase-2 design artifact. This document maps out an error handling strategy and rate-limiting constraints for each step of your distilled workflow.