AB Tasty Agent Skills

A collection of AI agent skills for integrating with the AB Tasty platform — covering Feature Experimentation & Rollouts (server-side) and Web Experimentation (client-side).

What Are Skills?

Skills are modular knowledge packages that extend AI coding agents (such as GitHub Copilot) with domain-specific expertise. Each skill exposes a SKILL.md entrypoint and a set of reference documents that guide the agent through complex, multi-step workflows.

Available Skill

abtasty

Expert guidance for AB Tasty's two main platforms:

Platform	Type	Description
Feature Experimentation & Rollouts	Server-side	Feature flags, rollouts, and server-side A/B tests
Web Experimentation	Client-side	Website A/B tests, personalizations, and modifications

Entrypoint: SKILL.md

Repository Structure

skills/
├── SKILL.md                        # Skill definition and routing logic
├── README.md                       # This file
├── LICENSE.txt
├── references/                     # Detailed reference documents
│   ├── decision-api.md             # Runtime feature flag retrieval (Decision API)
│   ├── resource-loader.md          # Campaign configuration loader
│   ├── fear-resource-extractor.md  # Feature Experimentation campaign generator
│   ├── we-resource-extractor.md    # Web Experimentation campaign generator
│   ├── nodejs-sdk.md               # Node.js SDK installation
│   ├── react-sdk.md                # React SDK installation
│   ├── python-sdk.md               # Python SDK installation
│   ├── java-sdk.md                 # Java SDK installation
│   ├── go-sdk.md                   # Go SDK installation
│   ├── php-sdk.md                  # PHP SDK installation
│   ├── dotnet-sdk.md               # .NET SDK installation
│   ├── android-sdk.md              # Android SDK installation
│   ├── ios-sdk.md                  # iOS SDK installation
│   └── flutter-sdk.md              # Flutter SDK installation
└── scripts/
    └── call-decision-api.js        # Working example for the Decision API

Reference Documents

SDKs

File	Platform	Language / Runtime
nodejs-sdk.md	Feature Experimentation	Node.js
react-sdk.md	Feature Experimentation	React / JavaScript
python-sdk.md	Feature Experimentation	Python 3.7+
java-sdk.md	Feature Experimentation	Java 8+
go-sdk.md	Feature Experimentation	Go 1.18+
php-sdk.md	Feature Experimentation	PHP 7.4+
dotnet-sdk.md	Feature Experimentation	.NET 6+
android-sdk.md	Feature Experimentation	Android (Kotlin / Java)
ios-sdk.md	Feature Experimentation	iOS (Swift / Obj-C)
flutter-sdk.md	Feature Experimentation	Flutter / Dart

APIs & Tooling

File	Description
decision-api.md	Retrieve feature flags and campaigns at runtime for a given visitor
resource-loader.md	Load and apply campaign configurations via the Resource Loader API
fear-resource-extractor.md	Generate Feature Experimentation campaign configs from natural language
we-resource-extractor.md	Generate Web Experimentation campaign configs from natural language

Common Workflows

Install SDK and retrieve feature flags

1. Follow the SDK guide for your platform (see SDK table above)
2. Use decision-api.md to get flags for a visitor
3. Reference scripts/call-decision-api.js for a working example

Create and deploy a new campaign

1. Generate a campaign config from a description using fear-resource-extractor.md (Feature Experimentation) or we-resource-extractor.md (Web Experimentation)
2. Load the resources with resource-loader.md

Credentials Reference

Credential	Where to find it
env_id / api_key	AB Tasty → Settings → Feature Experimentation → Environment settings
fear_rca_account_id / fear_rca_account_environment_id / fear_rca_token	Feature Experimentation account settings
we_account_id / we_token	Web Experimentation account settings

License

See LICENSE.txt.