Agentic Engineeringv1.0.0

A software development team adopts the eval-first approach to add a new feature. They define acceptance criteria, write capability and regression tests, decompose tasks into 15-minute units, and use model routing to optimize costs.

A team uses the skill to refactor a legacy codebase. They use Opus for architecture decisions, Sonnet for implementation, and Haiku for simple edits, while maintaining eval baselines to ensure no regressions.

A QA team leverages the skill to set up automated testing. They create baseline evals, use model routing for test generation, and implement a continuous evaluation loop to catch issues early.

A cloud-native team decomposes a monolithic application into microservices using the 15-minute unit rule. They define completion criteria for each service, use model routing for implementation, and run post-implementation evals.

An enterprise AI team uses the cost discipline aspects to track model usage per task, reduce spend by matching complexity to model tier, and document reasoning gaps before escalating.

A company provides a SaaS platform and uses the skill to accelerate feature delivery, reduce dev costs, and maintain high quality through evals, passing savings to customers or increasing margins.

Consultants offer services to help teams adopt agentic engineering practices, including eval-first workflows and model routing, charging for implementation support and training.

A tooling company builds plugins or platforms that integrate these workflow patterns, selling access to automation that enforces eval-first and cost-aware routing.