kvasir-approach-evaluation

Compare multiple approaches to a problem, evaluating trade-offs, feasibility, and alignment with constraints.

SKILL.md

GitHub

Approach Evaluation

Purpose

Systematically compare multiple candidate approaches to a problem, evaluating each against trade-offs, feasibility, constraints, and project goals. Recommend the approach best suited to the situation.

When to Use

  • When multiple valid approaches exist and a decision must be made.
  • When trade-offs between speed, quality, cost, or risk need to be weighed.
  • When constraints (time, team skill, tooling, architecture) eliminate some approaches.
  • Before committing to an implementation strategy.

Workflow

  1. Enumerate approaches.

    • Identify all viable candidate approaches, not just the obvious one.
    • Include approaches that differ meaningfully in strategy or effort.
    • Discard only those that are clearly infeasible, and note why.
  2. Define evaluation criteria.

    • Establish the dimensions of comparison relevant to this task.
    • Consider effort, risk, maintainability, scalability, and fit with existing patterns.
    • Ensure the criteria reflect the actual constraints and goals.
  3. Score each approach.

    • Evaluate each candidate consistently against every criterion.
    • Base scores on evidence and reasoning, not first impressions.
    • Record where an approach clearly excels or falls short.
  4. Identify trade-offs.

    • Make explicit what each approach gains and what it sacrifices.
    • Surface hidden costs such as complexity, lock-in, or maintenance burden.
    • Compare trade-offs against the priorities that matter most here.
  5. Recommend.

    • Select the approach best aligned with constraints and goals.
    • Justify the recommendation against the criteria and trade-offs.
    • State the conditions under which a different approach would be preferred.

Quality Criteria

  • All viable approaches are considered, not just the first that comes to mind.
  • Evaluation criteria are explicit and relevant to the task.
  • Trade-offs are concrete and specific.
  • The recommendation is justified, not asserted.

Anti-Patterns

  • False dichotomy: Presenting only two options when more exist.
  • Criteria mismatch: Evaluating against criteria irrelevant to the actual task.
  • Unjustified recommendation: Picking an approach without explaining why.
  • Ignoring constraints: Recommending an approach that violates known constraints.