technology-evaluation
Technology evaluation: fitness functions, landscape, proportional governance, migration guidance. Decide with trade-offs and reversibility.
Files1
View on GitHubSKILL.mdskills/research/technology-evaluation/SKILL.md
GitHubtechnology-evaluation
Decide with trade-offs and reversibility: fitness functions, proportional process, migration path.
Activate when
- Adopting frameworks, platforms, databases, or vendors; proposing migrations or deprecations.
Do NOT activate for
- Single-library picks (see library-selection); version lookups (see documentation-search).
Inspection
Gather requirements, constraints (team, compliance, scale, budget), current-state pain with evidence, and candidate landscape (docs, releases, community health, hiring implications).
Decision rules
- Fitness functions: 3–5 measurable criteria the choice must satisfy (latency, cost, operability, hiring, compliance) — scored, not vibed.
- Reversibility rules: easily reversible → decide fast, document briefly; hard to reverse (DB, cloud, framework) → spike first, ADR mandatory.
- Proportional governance: solo/small-team choices need a paragraph; org-wide bets need a reviewed ADR with migration + rollback plan.
- Total cost: license + operations + learning + exit cost. The cheapest license is rarely the cheapest choice.
Procedure
- Spike the top 2 candidates against the fitness functions (timeboxed).
- Record the ADR: context, options with scores, decision, migration plan, rollback plan, review date.
- Pilot on a bounded slice before full commitment.
- Verify: fitness functions pass on the pilot, rollback demonstrated or explicitly waived with owner.
Failure modes
- Resume-driven adoption; evaluation without a spike; ignoring exit cost; unbounded "evaluate everything" phases; ADR written to justify a foregone conclusion.
Escalation
Deep-dive research method → research/documentation-search; SaaS/vendor specifics → saas/multitenancy-billing, cloud/*.
References
- Related:
../library-selection/,../source-evaluation/,../../core/requirements-analysis/.