Core Concepts: Skills, Bundles, Workflows

Antigravity Awesome Skills provides a structured approach to empowering AI coding assistants with reusable, context-rich capabilities. This system is built upon three fundamental concepts: Skills, Bundles, and Workflows. Understanding these concepts is key to effectively leveraging the library for your agentic tasks.

Skills

At the heart of the Antigravity Awesome Skills library are Skills. A skill is a self-contained, reusable SKILL.md playbook designed to provide an AI agent with precise operating instructions for performing a specific, recurring task.

What they are

Think of a skill as an atomic unit of expertise for your AI assistant. Each skill is typically represented by a Markdown file (SKILL.md) that encapsulates:

Clear objectives and constraints for the task.
Detailed instructions on how to execute the task.
Examples of expected inputs and desired outputs.
Relevant context, best practices, or specific tools the agent should utilize.
(Optionally) Associated code assets or scripts that the skill might reference or use, making it a more comprehensive playbook.

Purpose

Skills move beyond one-off prompt engineering by providing a structured, robust, and repeatable way for AI agents to understand and perform tasks. They inject better context and stronger constraints, leading to more consistent and higher-quality outputs across a range of AI tools like Claude Code, Cursor, and Copilot.

Usage

When an AI assistant is given a skill, it gains the ability to apply that particular expertise to a problem. For example, a fix-security-vulnerability skill would equip an agent with the specific knowledge and steps required to address common security issues according to predefined guidelines.

Bundles

Bundles are curated collections of related Skills, designed to streamline the provisioning of capabilities to AI agents based on specific roles, domains, or project types.

What they are

Instead of individually selecting hundreds of skills, bundles allow you to install a cohesive set of skills tailored for a particular context. For instance, a "Frontend Developer Bundle" might include skills for React component creation, CSS debugging, API integration, and UI testing.

Purpose

Bundles serve as logical groupings that simplify discovery and installation. They ensure that an AI agent, when assigned a particular role or task area, has immediate access to a comprehensive set of relevant skills without tedious manual curation.

Usage

Users can choose to install a broad multi-tool skill library or opt for specific bundles to quickly equip their agents with a predefined set of expertise. This is often done via the npx installer:

npx antigravity-awesome-skills install --bundle devops

Workflows

Workflows represent an orchestration layer that combines multiple Skills into a sequential or conditional execution path to achieve more complex, multi-step objectives.

What they are

Workflows define a sequence of operations where the output of one skill might become the input for another. They are blueprints for how AI agents can tackle larger projects, breaking them down into manageable, interconnected steps. Examples include "Planning a new feature," "End-to-end testing a microservice," or "Performing a comprehensive security review."

Purpose

Workflows enable AI agents to perform sophisticated tasks that require logical progression and decision-making across various stages. They transform individual skills into powerful, automated processes, allowing agents to handle tasks like planning, coding, debugging, testing, and deployment without constant human intervention.

Usage

Workflows provide a higher-level abstraction for directing agent behavior. Instead of telling an agent "write code," a workflow might guide it through a series of steps such as "plan the module -> write the tests -> implement the feature -> refactor the code -> review for security," leveraging distinct skills at each step to ensure comprehensive execution.