Skip to content

[FEAT] Generate Pugh Decision Matrix from ADRs via LLM #28

Description

@divideby0

Summary

Use an LLM to generate a structured Pugh decision matrix from an Architecture Decision Record (ADR) as input. The LLM should extract or infer:

  • Options being compared (e.g., Zustand vs. Redux vs. Jotai)
  • Criteria relevant to the decision (e.g., bundle size, learning curve, dev experience)
  • Weights for each criterion based on context in the ADR
  • Ratings for each option against each criterion

Motivation

ADRs already capture the reasoning behind architectural decisions, but translating that into a structured comparison matrix is manual and tedious. LLM-assisted generation lets teams get a first-draft Pugh matrix instantly, which they can then refine.

Proof of Concept

Eli is building an initial POC using the Zustand evaluation ADR as the test case. This issue captures the full feature scope beyond that POC.

Considerations

  • Input format: markdown ADR (could support multiple ADR templates)
  • Output: a valid Pugh matrix data structure compatible with the existing component
  • Should support iterative refinement — user can adjust the generated matrix after creation
  • Prompt engineering and model selection TBD

Related

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions