Files
magnus919_agent-skills/software-architecture-analysis/README.md
T
Magnus HedemarkandGitHub db2ad850be feat(architecture): deepen codebase analysis (#355)
Add evidence lenses for characteristics, coupling, decomposition, data authority, workflows, health reporting, and eval coverage.\n\nAI-assisted: Jasper orchestrated implementation and verification with OpenCode.

Signed-off-by: Magnus Hedemark <magnus919@pm.me>
2026-08-21 00:09:03 -04:00

39 lines
2.5 KiB
Markdown

# Software Architecture Analysis — Reverse Engineering to Design Document
Reverse-engineer an existing codebase, assess its architecture health, and produce a clean-room design document, PRD, or migration plan grounded in repository evidence.
## Why Install This Skill
When your agent loads this skill, it becomes a **codebase archaeologist** who can:
- **Map repository structure** — identify core components, languages, and frameworks
- **Extract architecture** — understand how the system is actually built, not how it's documented
- **Inventory features** — catalog every capability the system provides
- **Identify implicit contracts** — storage operations, data flows, integration points
- **Assess architecture health** — quality characteristics, six coupling lenses, modularity, data authority, workflow failure, and reconciliation
- **Test decomposition readiness** — compare a bounded split with retaining a modular monolith instead of assuming services are better
- **Design clean-room alternatives** — re-imagine the system under new constraints (local-first, privacy-first, self-hosted)
- **Produce specifications** — PRDs, design documents, migration plans with zero source code copying
## What You Get
| Directory | Purpose |
|-----------|---------|
| `SKILL.md` | 7-phase build workflow, trigger conditions |
| `references/` | Interface extraction, quality-characteristic evidence, coupling/decomposition, and data/workflow analysis |
| `templates/` | Architecture health assessment worksheet |
| `evals/` | Output-quality cases for evidence, boundaries, decomposition, and distributed workflows |
## Triggers
Load this when you need to understand how an existing codebase works, produce a design document from implementation evidence, assess architecture health, map data ownership and distributed workflows, or evaluate readiness for a boundary change. Do not use it for greenfield architecture, direct code review, implementation, security auditing, API contract authoring, data-platform strategy, or migration execution.
## Requirements
Git and a programming language runtime matching the target codebase. Mermaid-capable Markdown is useful for architecture diagrams. No purchased books or private source material are required.
## Quick Start
Start with the workflow in `SKILL.md`, then load the relevant reference. For a health review, copy `templates/architecture-health-assessment.md`, fill the evidence ledger first, and keep recommendations in a separate section.