agentleFS
jasonkneenjasonkneen/kiro/skills/requirements-engineering/SKILL.md

requirements-engineering

Skill753 starsChanged 11 months agoMITView on GitHub

Transform vague feature ideas into clear, testable requirements using EARS format. Capture user stories, define acceptance criteria, identify edge cases, and validate completeness before moving to design.

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use requirements engineering when:

    line 19
  2. Distinctive:

    Bad: Only documenting happy path

    line 167
  3. Distinctive:

    Use requirements as foundation for acceptance testing

    line 266

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Requirements Engineering

Master the art of capturing what needs to be built before diving into how to build it. This skill teaches the EARS (Easy Approach to Requirements Syntax) format for creating clear, testable requirements.

When to Use This Skill

Use requirements engineering when:

  • Starting any new feature or project
  • Clarifying ambiguous stakeholder requests
  • Creating acceptance criteria for user stories
  • Documenting system behavior for testing
  • Ensuring all team members share understanding

The EARS Format

EARS provides consistent patterns for writing requirements that are specific, testable, and unambiguous.

Basic Patterns

Event-Response (Most Common):

WHEN [triggering event] THEN [system] SHALL [required response]

Conditional Behavior:

IF [precondition is met] THEN [system] SHALL [required response]

Complex Conditions:

WHEN [event] AND [additional condition] THEN [system] SHALL [response]

Optional Conditions:

WHEN [event] OR [alternative event] THEN [system] SHALL [response]

Advanced Patterns

State-Based:

WHEN [system is in specific state] THEN [system] SHALL [behavior]

Performance:

WHEN [user action] THEN [system] SHALL [respond within X seconds/milliseconds]

Security:

IF [authentication condition] THEN [system] SHALL [security response]

Step-by-Step Process

Step 1: Capture User Stories

Format: As a [role], I want [feature], so that [benefit]

Focus on:

  • Who is the user? (role)
  • What do they want to accomplish? (feature)
  • Why does it matter? (benefit/value)

Example:

As a returning customer, I want to save my payment methods, so that I can checkout faster in the future.

Step 2: Generate Acceptance Criteria

For each user story, define specific acceptance criteria using EARS:

Example for payment methods:

**User Story:** As a returning customer, I want to save my payment methods, so that I can checkout faster.

**Acceptance Criteria:**
1. WHEN user adds a valid credit card THEN system SHALL securely store card details
2. WHEN user adds a card with invalid number THEN system SHALL display validation error
3. WHEN user has saved cards THEN system SHALL display list during checkout
4. WHEN user selects saved card THEN system SHALL pre-fill payment form
5. WHEN user deletes saved card THEN system SHALL remove card from list
6. IF user is not authenticated THEN system SHALL redirect to login before saving card
7. WHEN user adds card THEN system SHALL mask all but last 4 digits in display

Step 3: Identify Edge Cases

For each requirement, ask:

  • What if the input is empty/null?
  • What if the input is at boundary values?
  • What if the operation fails?
  • What if the user is not authorized?
  • What if there are concurrent operations?

Edge case patterns:

**Error Handling:**
- WHEN [operation fails] THEN system SHALL [display error / retry / log]

**Boundary Conditions:**
- WHEN [value equals minimum/maximum] THEN system SHALL [specific behavior]

**Concurrent Access:**
- WHEN [multiple users access same resource] THEN system SHALL [conflict resolution]

**Empty States:**
- WHEN [collection is empty] THEN system SHALL [display empty state message]

Step 4: Validate Requirements

Use this checklist:

Completeness:

  • All user roles identified and addressed
  • Normal flow scenarios covered
  • Edge cases documented
  • Error cases handled
  • Business rules captured

Clarity:

  • Each requirement uses precise language
  • No ambiguous terms (fast, easy, user-friendly)
  • Technical jargon avoided or defined
  • Expected behaviors are specific

Consistency:

  • EARS format used throughout
  • Terminology consistent across requirements
  • No contradictory requirements
  • Similar scenarios handled similarly

Testability:

  • Each requirement can be verified
  • Success criteria are observable
  • Inputs and expected outputs specified
  • Performance requirements are measurable

Common Mistakes to Avoid

Mistake 1: Vague Requirements

Bad: "System should be fast" Good: "WHEN user submits search THEN system SHALL return results within 2 seconds"

Mistake 2: Implementation Details

Bad: "System shall use Redis for caching" Good: "WHEN user requests frequently accessed data THEN system SHALL return cached results"

Mistake 3: Missing Error Cases

Bad: Only documenting happy path Good: Include WHEN/IF statements for all error conditions

Mistake 4: Untestable Requirements

Bad: "System should be user-friendly" Good: "WHEN new user completes onboarding THEN system SHALL require no more than 3 clicks to reach main dashboard"

Mistake 5: Conflicting Requirements

Bad: Requirements that contradict each other Good: Review all requirements together, resolve conflicts explicitly

Examples

Example 1: File Upload Feature

**User Story:** As a user, I want to upload files, so that I can share documents with my team.

**Acceptance Criteria:**
1. WHEN user selects file under 10MB THEN system SHALL accept file for upload
2. WHEN user selects file over 10MB THEN system SHALL display "file too large (max 10MB)" error
3. WHEN user selects unsupported file type THEN system SHALL display "unsupported format" error with list of allowed types
4. WHEN upload is in progress THEN system SHALL display progress indicator with percentage
5. WHEN upload completes successfully THEN system SHALL display success message with file link
6. WHEN upload fails due to network error THEN system SHALL display retry option
7. IF user is not authenticated THEN system SHALL redirect to login before upload
8. WHEN user uploads file with same name as existing file THEN system SHALL prompt for rename or replace

**Supported File Types:** PDF, DOC, DOCX, XLS, XLSX, PNG, JPG, GIF
**Maximum File Size:** 10MB
**Maximum Files Per Upload:** 5

Example 2: Search Feature

**User Story:** As a customer, I want to search products, so that I can find items quickly.

**Acceptance Criteria:**
1. WHEN user enters search term THEN system SHALL display matching products
2. WHEN search returns results THEN system SHALL show result count
3. WHEN search returns no results THEN system SHALL display "no products found" with suggestions
4. WHEN user searches with special characters THEN system SHALL sanitize input and search
5. WHEN user submits empty search THEN system SHALL display validation message
6. WHEN results exceed 20 items THEN system SHALL paginate with 20 items per page
7. WHEN user searches THEN system SHALL return results within 2 seconds
8. WHEN user types in search box THEN system SHALL show autocomplete suggestions after 3 characters

**Search Fields:** Product name, description, category, SKU
**Minimum Search Length:** 2 characters

Requirements Document Template

# Requirements Document: [Feature Name]

## Overview
[Brief description of the feature and its purpose]

## User Roles
- [Role 1]: [Description of this user type]
- [Role 2]: [Description of this user type]

## Requirements

### Requirement 1: [Name]
**User Story:** As a [role], I want [feature], so that [benefit]

**Acceptance Criteria:**
1. WHEN [event] THEN system SHALL [response]
2. IF [condition] THEN system SHALL [response]
3. WHEN [event] AND [condition] THEN system SHALL [response]

**Edge Cases:**
- [Edge case 1 and how it's handled]
- [Edge case 2 and how it's handled]

### Requirement 2: [Name]
[Continue pattern...]

## Non-Functional Requirements
- **Performance:** [Specific metrics]
- **Security:** [Security requirements]
- **Accessibility:** [Accessibility standards]

## Out of Scope
- [Items explicitly not included in this feature]

## Open Questions
- [Questions that need stakeholder input]

Next Steps

After completing requirements:

  1. Review with stakeholders for accuracy
  2. Get explicit approval before proceeding
  3. Move to Design Phase to create technical architecture
  4. Use requirements as foundation for acceptance testing

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.