Define Prompt Skill

This skill is an action router — pick the step that matches the user's intent and execute only that step. Do not run other steps; do not parallelize.

Available actions:

• Step 1 — Single-string systemPrompt (default — when a one-paragraph instruction is enough)
• Step 2 — prompt { ... } builder with few-shot examples and structured turns
• Step 3 — Runtime augmentation via PromptAugmenter

Step 1 — Single-String systemPrompt

If the user just wants to set an instruction string, the factory's systemPrompt parameter is the right surface. No DSL needed:

val agent = AIAgent(
    promptExecutor = ...,
    llmModel = ...,
    systemPrompt = """
        You are a GitHub triage assistant. Classify issues, suggest labels,
        and link related issues by number when relevant.
    """.trimIndent(),
)

If the user is reaching for more structure than a single string supports, escalate to Step 2.

Finish here.

Step 2 — prompt { ... } Builder

Use the DSL when you need system + assistant + user turns interleaved (few-shot examples), or when the same prompt shape is reused across multiple agents.

import ai.koog.prompt.dsl.prompt

val triagePrompt = prompt("issue-triage") {
    system("You are a GitHub triage assistant. Classify issues as bug, feature, or question.")

    // Few-shot example 1
    user("App crashes on Windows when I open the Settings dialog.")
    assistant("""{"classification": "bug", "confidence": 0.95}""")

    // Few-shot example 2
    user("Could we add dark mode to the export view?")
    assistant("""{"classification": "feature", "confidence": 0.9}""")

    // The actual turn is appended at runtime by the agent
}

The Prompt class lives in ai.koog.prompt.Prompt as of 1.0 (moved from ai.koog.prompt.dsl.Prompt, #2022); the prompt { ... } builder DSL stays in ai.koog.prompt.dsl.

Pass the built Prompt into the agent's strategy or context — the exact wiring point depends on whether the prompt is the agent's system prompt or a per-node call. For per-node use inside a strategy, send the prompt through llm.writeSession { appendPrompt { ... } }.

Finish here.

Step 3 — Runtime Augmentation via PromptAugmenter

When the prompt needs runtime data (user metadata, current time, retrieved facts), use the augmenter family rather than string-templating into a systemPrompt. The augmenter appends a MessagePart to an existing Message — message-level rewrites from earlier versions are gone.

Variants:

• SystemPromptAugmenter — append to the system message
• UserPromptAugmenter — append to the next user turn
• AgentcorePromptAugmenter — append for Bedrock AgentCore flows

import ai.koog.prompt.processor.SystemPromptAugmenter

class TimestampAugmenter : SystemPromptAugmenter {
    override fun augment(): MessagePart {
        return MessagePart.text("Current time is ${java.time.Instant.now()}.")
    }
}

// Install via the agent's prompt processor surface (see prompt-processor module)

Augmenters run on every agent invocation — keep them cheap. Heavy fact retrieval belongs in LongTermMemory's search pipeline (see manage-state), not in an augmenter.

Finish here.