user-intent-parser

name	user-intent-parser
description	Parse user requirements into structured format with explicit assumptions, constraints, and acceptance criteria. Use when initial requirements are ambiguous or informal.

User Intent Parser Skill

Purpose

Convert informal user requests into structured requirement format.

When to Use

• User provides vague or informal request
• Requirements are conversational rather than structured
• Non-technical user is specifying features
• Need to formalize verbal requirements

Parsing Process

Step 1: Extract Explicit Statements

Identify what user directly stated:

• Actions (verbs): create, update, delete, show, send, etc.
• Objects (nouns): user, product, order, notification, etc.
• Conditions (when/if): triggers, prerequisites
• Outcomes (so that): expected results

Step 2: Identify Implicit Requirements

What's assumed but not stated:

• Authentication required?
• Error handling expectations
• Performance expectations
• Platform/device support
• Data validation needs

Step 3: Flag Ambiguities

Mark unclear items:

• Vague terms ("fast", "good", "easy", "simple")
• Missing specifics (quantities, limits)
• Unclear scope (boundaries)
• Undefined actors (who does what)

Step 4: Generate Structured Format

Use templates/parsed-intent.md

Output Format

Parsed Intent Document

Save to: docs/specs/parsed-intent-{session}.md

The template captures:

• Original user statement
• Extracted functional requirements
• Extracted non-functional requirements
• Assumptions made (with rationale)
• Ambiguities requiring clarification
• Draft acceptance criteria

Confidence Levels

Assign confidence to each extracted requirement:

Level	Meaning	Action
High	Directly stated by user	Proceed
Medium	Strongly implied	Confirm
Low	Inferred/assumed	Must clarify

Common Patterns

Feature Requests

User: "I need users to be able to export their data"

Parsed:
- Action: export
- Object: user data
- Actor: users (authenticated)
- Implicit: format unspecified, permissions assumed
- Ambiguity: which data? what format?

Bug Reports as Features

User: "The search is too slow"

Parsed:
- Action: improve search performance
- Implicit: current performance is unacceptable
- Ambiguity: how slow? target speed?

Vague Requests

User: "Make the dashboard better"

Parsed:
- Action: improve dashboard
- Ambiguity: which aspects? visual? functional? performance?
- Confidence: Low (requires extensive clarification)

Integration Points

1. Research findings inform implicit requirements
2. Generated ambiguities feed into question generation
3. Structured output becomes input for requirements validation

Storage Location

Save to: docs/specs/parsed-intent-{session}.md

Quality Checklist

Before completing parsing:

• All explicit actions identified
• All objects/entities named
• Implicit requirements documented
• Ambiguities clearly flagged
• Confidence levels assigned
• Draft acceptance criteria created