directory-management

Creates project directory structures and organizes artifacts for SageMaker AI workflows.

881|152|Updated Feb 5, 2026
One-click install
npx skills add https://github.com/awslabs/agent-plugins --skill directory-management
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: directory-management
Source: https://github.com/awslabs/agent-plugins/tree/main/plugins/sagemaker-ai/skills/directory-management
Command: npx skills add https://github.com/awslabs/agent-plugins --skill directory-management

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

When starting or resuming machine learning projects, files often end up scattered and inconsistently named, making it hard to track plans, scripts, notebooks, and outputs. This Skill standardizes project setup by resolving a project name and creating a consistent folder structure for all generated artifacts.

Core Features & Use Cases

  • Project Name Resolution: Detects existing projects by scanning for */PLAN.md files, or recommends a lowercase slug (max 64 chars, [a-z0-9-]) and waits for user confirmation.
  • Standard Directory Scaffolding: Creates specs/, scripts/, notebooks/, manifests/, and agent_memory/ folders under the project directory.
  • Artifact Organization: Associates PLAN.md files, generated Python scripts, Jupyter notebooks, JSON manifests, and session notes with the correct project folder.
  • Use Case: When starting a new SageMaker customization experiment, ask the agent to set up the project so every generated script, notebook, and manifest lands in a predictable location, and resuming later reloads the existing PLAN.md.

Quick Start

Set up a new project directory for my SageMaker training experiment and organize all generated artifacts under it.

Frequently Asked Questions about directory-management

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

FAQPage Schema
How do I set up a project directory structure for an ML experiment?

Invoke the directory-management skill at the start of your work. It resolves a project name, then creates specs/, scripts/, notebooks/, manifests/, and agent_memory/ folders under a directory named after your project.

How do I resume an existing project with this skill?

The skill scans the current directory for existing */PLAN.md files. If found, it asks whether you are resuming that project and loads the PLAN.md into context before continuing work.

What naming rules apply to project directory names?

Project names must be lowercase slugs of at most 64 characters using only letters, digits, and hyphens ([a-z0-9-]). The skill recommends a name from conversation context and waits for your confirmation.

Where are generated scripts and notebooks stored?

Generated Python scripts go in scripts/ (e.g., <project-name>_transform_fn.py) and Jupyter notebooks in notebooks/ (e.g., <project-name>.ipynb), both under the project directory. Machine-readable JSON outputs go in manifests/.

What is the agent_memory folder used for?

The agent_memory/ directory stores session persistence files such as session-notes.md, which tracks progress, artifacts, and next steps. It is intended to be git-ignored so session state stays local.