CARE Framework for AI-Driven Code Development

The SoftwareSeni Specification Template suggests using the CARE framework to prevent AI from making arbitrary architectural choices:

• Context: Why this exists and where it fits in the architecture.
• Action: Detailed functional steps with few-shot examples.
• Result: Measurable acceptance criteria (e.g., “response < 200ms”).
• Evaluation: Non-functional constraints like security and error handling.

Example: AI-Ready Requirement

A typical vague requirement like “Handle user authentication” often leads to security vulnerabilities. Below is an example of an AI-Ready Requirement that produces high-quality code on the first try:

Section	AI-Ready Specification
User Story	As a user, I want to log in securely using my email and password.
Technical Context	Implement using JWT tokens with bcrypt password hashing.
Inputs/Data	email: String, max 255 chars, unique. password: String, min 8 chars, must include upper/lower/number/symbol.
Acceptance Criteria	1. Lock account for 10 mins after 3 failed attempts.
	2. Terminate session after 30 mins of inactivity.
	3. Return generic error messages to prevent user enumeration.
Edge Cases	Handle null inputs empty strings, and SQL injection patterns in name fields.
Success Examples	1. login(‘user@test.com’, ‘ValidPass123!’) → Returns 200 OK + JWT.
	2. login(‘wrong@test.com’, ‘pass’) → Returns 401 Unauthorized.

Example