chidekina/aria-superpowers/prompt-modeler

Prompt Modeler

Overview

Generate structured, high-quality prompts based on contextual project analysis.

This skill:

• Analyzes scope and constraints
• Classifies risk and rigor
• Selects appropriate strategic profiles
• Generates 3 prompt options
• Never executes the project
• Never assumes behavior of generated prompts

Use when transforming ideas, projects, or existing code into structured prompts.

Usage

Manual invocation:

/prompt-modeler

Execution always begins with the Mandatory Gate (Stage 0).

Workflow

Stage 0 — Mandatory Gate

Always ask:

Choose: (A) Describe the project (B) Analyze existing files

If describing, require:

• Main objective
• Type (Study / Portfolio / Freelance / Product / Experiment)
• Stack or technology
• Deadline (if any)
• Dominant focus (Learning / Delivery / Architecture / Automation)
• Rigor level (Low / Moderate / High)
• Allow code generation? (Yes / Warn before / No)
• Governance model (Continuous flow / Always confirm)

Do not proceed without this stage.

Stage 1 — Contextual Analysis

1. Extract dominant keywords
2. Classify:

• Project type
• Risk level (Low / Medium / High)
• Required rigor
• Dominant focus
• Scope ambiguity

1. Detect:

• Internal conflicts
• Organizational complexity (stakeholders, compliance, distributed architecture, enterprise context)

If input is insufficient:

• Ask up to 3 refinement questions

or

• Request improved description

Never generate prompts from weak input.

Strategic Profile Selection

Select up to 3 profiles.

Available profiles:

• Architect / Tech Lead
• Productive Copilot / Executor
• Senior Tutor
• Critical Reviewer
• Security Auditor
• Performance Specialist
• Technical Writer
• Product Strategist
• Organizational Architect (enterprise contexts only)

Priority rule (immutable):

Focus > Risk > Type > Rigor

Light Mode Criteria

Apply only if:

• Low risk
• Closed scope
• No organizational complexity

Light Mode keeps:

• Clear objective
• Explicit context
• Defined output format
• Quality criteria
• Final directive

Stage 2 — Mandatory Output Structure

Always return:

Diagnostic (max 4 lines)

• Type
• Focus
• Risk
• Organizational complexity (if any)

Primary Recommendation

• Dominant profile
• Short justification
• Confidence level (High / Medium / Low)

Option 1 Option 2 Option 3

Each option must include:

• Name
• When to use (1 line)
• Ready-to-copy prompt (separate block)

Custom Prompt Adjustment

• Direct refinement questions

Always end with:

Choose: 1, 2, 3 or Customize.

Constraints

• Never execute the project.
• Never assume behavior of generated prompts.
• Never generate fewer than 3 options.
• Maintain deterministic reasoning.
• Maintain profile priority rule.