architecture-tradeoff-analysis
Make the trade-off explicit for every architecture choice. The first law of software
SKILL.mdsubagents/software-architecture/skills/architecture-tradeoff-analysis/SKILL.md
GitHubArchitecture trade-off analysis
Purpose
Make the trade-off explicit for every architecture choice. The first law of software architecture is that everything is a trade-off: any structural option buys some properties at the expense of others, so the honest default answer to "what is the right architecture?" is "it depends on the driving forces." This skill turns that stance into a repeatable analysis that names what an option gains, what it sacrifices, and which forces decide.
When to use
- The caller proposes or reviews any structural choice where one option could be taken over another.
- The caller asks for the single best architecture, pattern, or technology, or presents an option's benefits with no cost stated.
- A recommendation is being written and needs its residual consequences surfaced before it ships.
Do not invoke for a settled factual definition that carries no design choice.
Procedure
- Restate the decision and the alternatives. Name the concrete options on the table (including the status quo / do-nothing). A trade-off analysis needs at least two options to compare; if only one is offered, surface the implied alternative.
- Surface the driving forces. Pull the prioritized architecture characteristics and
constraints the decision must serve. If they are unstated, that is the first gap — hand off to
identify-architecture-characteristicsbefore judging options. - For each option, state gain AND sacrifice. Write both halves explicitly: the property it improves and the property, cost, or flexibility it gives up. An option with only upside listed is an incomplete analysis, not a winner.
- Map each option to the prioritized forces. Score how well it serves the top-ranked characteristics, not every quality at once. The option that best serves the highest-priority forces wins even if it loses on lower-priority ones.
- Name the residual consequence. State plainly what the caller must accept by choosing the recommended option — the sacrifice does not disappear, it becomes a known cost.
- Recommend, tied to forces. Give one recommendation, justified by the ranked forces, never as a universal "best." If the forces genuinely do not separate the options, say the choice is indifferent and name the tie-breaker.
Inputs
- The decision under review and its candidate options.
- The system's driving forces: the quality attributes being optimised and the hard constraints.
Output
A trade-off table or narrative in which every option carries an explicit gain-versus-sacrifice pair, a recommendation tied to the prioritized forces, and the residual consequence the caller must accept. Never one option presented as universally best.
References
- Laws of software architecture — the first law (everything is a trade-off) and the trade-off-analysis frame this skill operationalizes.
Provenance
Distilled from principle(s) P001, claims C00001/C00002, evidence E00001/E00002, anchored in the cited architecture sources (sources/anchors/). Source is distillation-only: content is paraphrased, never quoted verbatim.