Operationalize LLM engineering with model routing
agentic-engineeringskillsetup L2★0
Sheshiyer/skill-clusters ↗What it does
Execute LLM engineering with eval-first
Best for
Engineering workflows where AI agents perform most implementation and evals enforce quality gates.
Inputs
- · Completion criteria
- · Implementation task
- · Baseline evals
Outputs
- · Agent-executed implementation
- · Eval comparisons
- · Regression report
Preconditions
Define evals before execution; decompose into 15-minute units
Failure modes
No evals = no regression signal; coupling between units = slow iteration
Trust signals
- · Eval-first discipline
- · Model-tier routing by complexity
- · Regression baseline checks