CtrlK
BlogDocsLog inGet started
Tessl Logo

jbaruch/koog

Koog 1.0 idioms, gotchas, and scaffolding skills for Kotlin agents on the JVM

86

1.86x
Quality

88%

Does it follow best practices?

Impact

86%

1.86x

Average score across 45 eval scenarios

SecuritybySnyk

Advisory

Suggest reviewing before use

Overview
Quality
Evals
Security
Files

criteria.jsonevals/scenario-43/

{
  "context": "Tests whether the agent reaches for the event handler feature for human-readable per-step output — the right tool for development/demo surface — rather than reaching for OpenTelemetry (production signal, overkill here) or hand-rolling println scaffolding inside tool bodies.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Installs the event-handler feature",
      "description": "Calls handleEvents { ... } inside the AIAgent(...) trailing lambda. Does NOT install OpenTelemetry (overkill for stdout demo output and adds collector setup) and does NOT add println calls inside each tool body",
      "max_score": 30
    },
    {
      "name": "Adds the agents-features-event-handler dependency",
      "description": "Adds ai.koog:agents-features-event-handler to build.gradle.kts at 1.0.0 or later. The umbrella does not include the event handler",
      "max_score": 20
    },
    {
      "name": "Subscribes to onToolCallStarting",
      "description": "Registers an onToolCallStarting handler that prints tool name and arguments. The start-side callback is essential for the live demo trace",
      "max_score": 20
    },
    {
      "name": "Subscribes to onToolCallFinished",
      "description": "Registers an onToolCallFinished handler that prints the tool result (or at minimum a completion marker). The task requires surfacing when each tool completes — both start and finish callbacks are needed",
      "max_score": 15
    },
    {
      "name": "Uses distinct visual markers for start vs end",
      "description": "Uses visually distinct prefixes between start and end output (e.g., → vs ←, or [start] vs [end]) so the audience can read the trace at scroll speed",
      "max_score": 10
    },
    {
      "name": "Does not also install OpenTelemetry",
      "description": "Does not add OpenTelemetry on top — the developer explicitly said no Langfuse/OTel for this demo. Adding it would require collector setup and isn't requested",
      "max_score": 5
    }
  ]
}

evals

README.md

tile.json