AI Skills

Plyson is built to be AI-native. Every new project includes a skills directory containing specialized "knowledge packs" that teach AI models how to write high-quality Plyson tests.

The General Principle

Regardless of which AI tool you use (Gemini CLI, Copilot, Cursor, Claude Code, ChatGPT, etc.), the workflow is the same:

1. Provide Context: Feed the contents of the SKILL.md files in your skills/ directory to the AI.
2. Assign the Task: Ask the AI to follow the instructions in those files to process your requirements.

How to use with your favorite tool

1. File-Based Assistants (IDE & CLI)

Tools that have access to your workspace (like Cursor, GitHub Copilot, Claude Code, or Windsurf) work best when you point them directly to the skill files.

• Prompt Pattern: "Using the logic in skills/sdet-testcase-generator/SKILL.md, generate test cases for..."
• Context Tip: Use @file, #file, or simply open the file in your editor so the AI can reference it.

2. Specialized CLI Agents

Tools like Gemini CLI often have dedicated folders for skills.

• Setup: Copy the skills to the tool's expected location (e.g., .gemini/skills/).
• Prompt Pattern: "Use the sdet-json-generator skill to..."

3. Web Interfaces (ChatGPT / Claude.ai)

If you are using a web-based chat, you can manually provide the context.

• Step: Copy the text from a SKILL.md file and paste it into the chat.
• Prompt Pattern: "I'm going to provide you with a 'Skill' definition. Please follow its instructions to convert these requirements into Plyson JSON: [Paste Skill Content] [Paste Requirements]"

---

The Two-Step Workflow

We recommend a two-step approach for maximum accuracy:

Step 1: Requirements to Markdown Scenarios

Use the sdet-testcase-generator to transform raw user stories into structured markdown test cases. This step ensures the business logic is correct before writing any code.

Step 2: Scenarios to executable JSON

Use the sdet-json-generator to transform those markdown scenarios into valid Plyson JSON suites. This step handles the technical implementation, schema compliance, and formatting.

Why it works

By providing the SKILL.md files as context, you are:

• Standardizing Output: Ensuring the AI uses the correct HTTP methods, operators, and JSON structure.
• Improving Coverage: Forcing the AI to think about boundary cases and security as defined in the skill instructions.
• Reducing Hallucinations: Giving the AI a grounded "source of truth" for how Plyson works.