Koog 1.0 idioms, gotchas, and scaffolding skills for Kotlin agents on the JVM
88
88%
Does it follow best practices?
Impact
88%
1.95xAverage score across 43 eval scenarios
Passed
No known issues
{
"context": "Negative case — tests whether the agent recognizes that the developer's actual workload (one-shot text-to-text with no tools and no distinct phases) does not benefit from the subtask-pipeline pattern, and routes them to a simpler shape rather than building the pipeline because the developer asked for it by name.",
"type": "weighted_checklist",
"checklist": [
{
"name": "Names the absence of distinct phases as the disqualifier",
"description": "Identifies in prose that the workload has no phases that produce structurally different intermediate artifacts — it is one transformation. The pipeline pattern's benefit is the typed handoff between phases; with no handoff there is no benefit",
"max_score": 25
},
{
"name": "Redirects to the simpler shape",
"description": "Recommends the default agent shape — `AIAgent(systemPrompt = ..., promptExecutor = ..., llmModel = ...)` (the default `singleRunStrategy()` is implicit and usually omitted) — OR the functional-agent shape if some Kotlin logic is needed. Names one of these specifically. Does NOT build a contrived subgraphWithTask<String, String> pipeline just to satisfy the developer's request",
"max_score": 30
},
{
"name": "Does not produce subgraphWithTask code",
"description": "The output contains no subgraphWithTask, subgraphWithVerification, or strategy { ... } block. Building the pipeline pattern for a one-shot text transform is the failure mode this scenario tests",
"max_score": 20
},
{
"name": "Does not introduce typed handoff data classes",
"description": "The output contains no @Serializable @LLMDescription'd intermediate data classes. Inventing a fake intermediate type just to have a 'typed handoff' for a one-step transform is over-engineering",
"max_score": 15
},
{
"name": "Engages with the developer's framing without capitulating",
"description": "Acknowledges that the pattern is real and powerful — and explains the criterion that makes it fit: distinct phases with structurally different intermediate shapes. Does not just refuse without explaining when the pattern would fit",
"max_score": 10
}
]
}evals
scenario-1
scenario-2
scenario-3
scenario-4
scenario-5
scenario-6
scenario-7
scenario-8
scenario-9
scenario-10
scenario-11
scenario-12
scenario-13
scenario-14
scenario-15
scenario-16
scenario-17
scenario-18
scenario-19
scenario-20
scenario-21
scenario-22
scenario-23
scenario-24
scenario-25
scenario-26
scenario-27
scenario-28
scenario-29
scenario-30
scenario-31
scenario-32
scenario-33
scenario-34
scenario-35
scenario-36
scenario-37
scenario-38
scenario-39
scenario-40
scenario-41
scenario-42
scenario-43
skills
add-observability
add-persistence
add-rag
add-structured-output
add-token-budgeting
add-tool
cache-llm-calls
define-prompt
domain-model-subtask-pipeline
references
enable-prompt-caching
handle-agent-events
manage-state
migrate-from-0-x
model-planner-subtasks
persist-chat-history
query-sql-from-agent
scaffold-agent
snapshot-and-restore
test-koog-agents
trace-agent-internals
use-attachments
use-functional-agent
use-llm-node-variants
use-planner
wire-a2a
wire-acp-server
wire-ktor-server
wire-mcp-server
wire-spring-boot