Trade off analysis

Every software architecture decision your team makes follows the same broken process.

Someone proposes microservices. Someone else argues for the monolith. A third person suggests "modular monolith" because they read a blog post last Tuesday. Three hours of whiteboarding later, the most senior person in the room wins—not because their reasoning is stronger, but because everyone is tired.

The decision gets documented in a two-paragraph Confluence page that says "we decided to go with Option B" and nothing about why.

This Claude skill replaces that entire process with a structured, scored, defensible architecture analysis — generated in minutes from a single paragraph describing your decision.

Describe your architecture decision in plain English. One paragraph. That's all.

You get back a complete analysis document that would take a senior architect a full day to produce:

Scored Trade-off Matrix — 5–8 architecture characteristics (scalability, reliability, maintainability, security, cost, deployability) weighted to your priorities, with every option scored on a calibrated 1–5 scale. The math is visible. The reasoning is explicit. No hand-waving.

Named Tensions — The 3 places where your stated priorities conflict with each other. The tensions most teams discover 6 months into implementation? They're surfaced before you write a single line of code.

"What Would Have to Be True" — Borrowed from Roger Martin's strategy framework: for each option, the specific conditions that would make it the right choice. This reframes the debate from "which is best" to "which assumptions are we most confident in."

Hidden Cost Analysis — The costs your team will miss because they don't show up on architecture diagrams: Conway's Law effects on team structure, on-call burden shifts, cognitive load increases, and migration risks that compound over quarters.

Risk Profile — Every option scored across four dimensions: technical risk, organizational risk, timeline risk, and reversibility. Because an irreversible bad decision is categorically different from a reversible one.

Fitness Functions — Measurable signals to monitor after the decision ships. If the decision was wrong, these are the metrics that will tell you first.

ADR-Ready Summary — A clean Architecture Decision Record you can paste directly into your documentation. Context, decision, consequences, status — formatted and ready.

Example

Here's the exact prompt someone fed the skill

"We're a fintech startup with 12 engineers. We currently have a Django monolith handling payments, user accounts, and a lending product. We're growing fast and need to decide: should we extract the payment service into a microservice, refactor into a modular monolith, or keep the monolith and optimize? We process about 50K transactions per day and expect 10x growth in the next 18 months."

The output: 250 lines of structured analysis comparing all three options across 8 weighted characteristics. Three named tensions. Per-option risk profiles across four dimensions. A clear recommendation with confidence caveats and the conditions under which that recommendation changes. Fitness functions to track post-decision.

The kind of document that normally takes a senior architect a full working day — produced in minutes.

📄 The full example output is included in the download so you can judge the quality yourself.

What is included

SKILL.md file — 180 lines of structured architecture decision methodology. The engine behind the analysis.

Characteristics Taxonomy — 25 architecture characteristics organized across 4 categories (operational, structural, cross-cutting, 2026-era), with 10 named tensions and selection heuristics. This is the vocabulary the skill uses to evaluate your options.

Scoring Rubric — Anchor-based 1–5 scale with calibration examples. No more "I'd give it a 4 out of 5 for scalability" with zero justification. Every score is defensible.

README — Step-by-step installation for Claude Code, Cursor, VS Code, and Cowork.

Example Output — The complete fintech analysis. See exactly what to expect before you buy.

How to install

Upload the skill in Claude(Desktop/Web).

Settings --> Capabilities --> Add --> Upload the .zip file

Go back to chat

Describe any architecture decision in plain English.

That's it. No configuration. No coding. No API wrangling.

Works with Claude Pro, Claude Max, or any API key.

Who is this for

Senior developers who are tired of architecture decisions being made by whoever talks loudest in the room.

Software architects who want their trade-off analysis to be structured, quantified, and defensible — not a vibes-based whiteboard sketch that gets erased on Monday.

Engineering leads evaluating technology choices who need to communicate the why behind decisions to stakeholders who weren't in the room.

Anyone who has spent 3 hours in a meeting debating microservices vs. monolith and walked out without a clear framework for why the team chose what it chose.

Part of Architect OS (coming soon)

This is 1 of 45 AI-powered skills in the upcoming Architect OS — a complete operating system for software architects.

The full OS includes

• 45 skills across decisions, design, communication, analysis, governance, leadership, and strategy

• 8 sub-agents for cross-functional architecture review

• 9 frameworks wired directly into workflows

• A context library that calibrates every output to your company, your stack, and your constraints

Buy this skill now. When Architect OS launches, you'll already have the most important piece.