Elicit Requirements Models

Common Contract (Load First)

When this skill is activated, first load skill-common and apply the shared conventions it defines: canonical section order, frontmatter rules, anti-pattern guidance, and the Review In Mind loop in review-in-mind. This skill adds only domain-specific rules below.

Who

Business analysts, system architects, developers, and AI agents. The agent uses this skill to critique requirements for completeness.

What

This skill parses unstructured or flat text requirements and converts them into structured visual models (e.g., State Transition Diagrams, Data Flow Diagrams) using Mermaid.js syntax. It then analyzes the generated model to identify "missing arrows" (unhandled edge cases or missing requirements).

When

Invoke this skill during the requirements analysis phase, before architectural design or coding begins. Trigger phrases include: "model these requirements," "draw a state diagram for this," "are we missing any edge cases," or "visualize this user story."

Where

Applies to User Stories, Product Requirements Documents (PRDs), and business rule lists.

Why

Human language is inherently ambiguous and linear. When describing a system, stakeholders often only describe the "happy path." By forcing text into a strict visual model, dead-ends, unhandled errors, and missing transitions become immediately obvious.

Inputs

• Raw text requirements or user stories.
• Existing system context (optional).

Output (Logical Evidence)

• A Mermaid.js diagram block representing the logic.
• A list of "Missing Requirements" discovered by analyzing the model.

Optimization Readiness

• Failure Signals: The wrong model type is chosen, artifacts stay too abstract, clarifying questions are not surfaced, or the analysis drifts into solution design before the problem is understood.
• Evidence To Collect: Model drafts, missing-requirement lists, clarification questions, and examples showing which model type fit the input best.
• Safe Mutation Boundaries: Refine model-selection prompts, question-asking guidance, notation cues, and output framing without changing the core requirement-elicitation objective.
• Acceptance Criteria: Accept revisions only if the selected model matches the problem shape, unanswered questions are explicit, and the output remains understandable without additional explanation.
• Rejected Revision Handling: Record over-specific models, premature solutions, and weak question-handling patterns so they are not reused.
• Transfer Check: Verify the workflow still works across feature discovery, analysis, and clarification-heavy stakeholder sessions.
• Stop Rule: If the business domain or requirement signal is too unclear to model safely, stop and ask before choosing a technique.

Constraints (Logical Boundaries)

• Do not invent logic to fill gaps; explicitly highlight the gaps as questions for the stakeholder.
• Keep models scoped to the provided text to avoid diagram bloat.
• Keep instructions executable via natural language checklists; visual diagrams are optional outputs.
• If Mermaid cannot be rendered, provide equivalent state/flow structures in plain text form.

One More Thing

If the input text is purely cosmetic (e.g., "Change the button to blue") and contains no logic, state, or flow, inform the user that visual modeling is not applicable.

How (Structural Workflow)

Phase 1: Model Selection

1. Analyze the text to determine the dominant logic:
   • If it describes an object changing status over time -> State Transition Diagram.
   • If it describes a sequence of user/system actions -> Flowchart / Activity Diagram.
   • If it describes how the system interacts with external actors -> Context Diagram.

Phase 2: Translation to Visual Model

2. Extract the entities, states, actors, or steps from the text.
3. Draft the model using Mermaid.js syntax.

Phase 3: Gap Analysis

4. Visually trace the model to find gaps:
   • Are there states with no exit path?
   • Are there decisions (diamonds) with only one outgoing arrow?
   • What happens on failure/timeout at each step?
5. Generate a list of clarifying questions for the missing requirements.

Resources

• Mermaid Modeling Templates

Review In Mind (ReviewInMindGenie)

Execute the common review loop in review-in-mind before delivering.

Review lens for this skill:

• Does each generated model match the source requirements and reveal missing arrows/edge cases?
• Is the Mermaid syntax valid and the model readable by the intended audience?
• Are gaps reported as findings rather than silently filled?

Validation

1. Verify the Mermaid diagram renders correctly without syntax errors.
2. Ensure the gap analysis explicitly references specific missing transitions or unhandled states from the diagram.