cleandataengineering.aiAgent Harness, drawn simply
You are here: Tech stack › PS · Performance & scalability

Agent Harness Tutorial · Tech stack

PS · Performance & scalability

Fast enough at the 95th percentile; more servers when load rises; the model quota is the real limit. Family: Act · around every turn.

AZURE OPTIONSContainer Apps autoscaling✓ used: minimalAzure Load Testing◐ planned: recommendedManaged Redis· considered: optionalOUR CHOICEStateless Container Apps0 to 1 replicas in devPer-turn time budget60 sCache-friendly context orderOPEN-SOURCE OPTIONS(none named)· consideredWHO DOES THE WORKAzure gives 55% · we code once 35% · each agent team still does 10%
Dark boxes are what we run. Solid boxes are used or planned; dashed boxes were considered or dropped (crossed). The bar: grey is what Azure gives, dark is what we code once, hatched is what each agent team writes.
Why our choiceStateless servers scale out with no code change. The real ceiling is the model quota.
OptionWhere fromVerdictWhy
Stateless Container Appsours✓ chosen0 to 1 replicas in dev
Per-turn time budgetours✓ chosen60 s
Cache-friendly context orderours✓ chosen
Container Apps autoscalingAzure✓ usedminimal
Azure Load TestingAzure◐ plannedrecommended
Managed RedisAzure· consideredoptional
(none named)open source· considered

How it works inside the harness: Core loop.

How it is done: the real code

Cut from the live repositories each time the site is built, so it always matches what runs.

      scale: {
        minReplicas: 0
        maxReplicas: 1
        rules: [