subagent-driven-development
Dispatch fresh subagents per task with two-stage review (spec compliance, then code quality) in the current session.
kotlin-mcp-server-generator
Generate production-ready Kotlin MCP server projects with Gradle, official SDK, and typed tools.
Full skill instructions
Generate a complete, production-ready Model Context Protocol (MCP) server project in Kotlin.
You will create a Kotlin MCP server with:
myserver/
├── build.gradle.kts
├── settings.gradle.kts
├── gradle.properties
├── src/
│ ├── main/
│ │ └── kotlin/
│ │ └── com/example/myserver/
│ │ ├── Main.kt
│ │ ├── Server.kt
│ │ ├── config/
│ │ │ └── Config.kt
│ │ └── tools/
│ │ ├── Tool1.kt
│ │ └── Tool2.kt
│ └── test/
│ └── kotlin/
│ └── com/example/myserver/
│ └── ServerTest.kt
└── README.md
plugins {
kotlin("jvm") version "2.1.0"
kotlin("plugin.serialization") version "2.1.0"
application
}
group = "com.example"
version = "1.0.0"
repositories {
mavenCentral()
}
dependencies {
implementation("io.modelcontextprotocol:kotlin-sdk:0.7.2")
// Ktor for transports
implementation("io.ktor:ktor-server-netty:3.0.0")
implementation("io.ktor:ktor-client-cio:3.0.0")
// Serialization
implementation("org.jetbrains.kotlinx:kotlinx-serialization-json:1.7.3")
// Coroutines
implementation("org.jetbrains.kotlinx:kotlinx-coroutines-core:1.9.0")
// Logging
implementation("io.github.oshai:kotlin-logging-jvm:7.0.0")
implementation("ch.qos.logback:logback-classic:1.5.12")
// Testing
testImplementation(kotlin("test"))
testImplementation("org.jetbrains.kotlinx:kotlinx-coroutines-test:1.9.0")
}
application {
mainClass.set("com.example.myserver.MainKt")
}
tasks.test {
useJUnitPlatform()
}
kotlin {
jvmToolchain(17)
}
rootProject.name = "{{PROJECT_NAME}}"
package com.example.myserver
import io.modelcontextprotocol.kotlin.sdk.server.StdioServerTransport
import kotlinx.coroutines.runBlocking
import io.github.oshai.kotlinlogging.KotlinLogging
private val logger = KotlinLogging.logger {}
fun main() = runBlocking {
logger.info { "Starting MCP server..." }
val config = loadConfig()
val server = createServer(config)
// Use stdio transport
val transport = StdioServerTransport()
logger.info { "Server '${config.name}' v${config.version} ready" }
server.connect(transport)
}
package com.example.myserver
import io.modelcontextprotocol.kotlin.sdk.server.Server
import io.modelcontextprotocol.kotlin.sdk.server.ServerOptions
import io.modelcontextprotocol.kotlin.sdk.Implementation
import io.modelcontextprotocol.kotlin.sdk.ServerCapabilities
import com.example.myserver.tools.registerTools
fun createServer(config: Config): Server {
val server = Server(
serverInfo = Implementation(
name = config.name,
version = config.version
),
options = ServerOptions(
capabilities = ServerCapabilities(
tools = ServerCapabilities.Tools(),
resources = ServerCapabilities.Resources(
subscribe = true,
listChanged = true
),
prompts = ServerCapabilities.Prompts(listChanged = true)
)
)
) {
config.description
}
// Register all tools
server.registerTools()
return server
}
package com.example.myserver.config
import kotlinx.serialization.Serializable
@Serializable
data class Config(
val name: String = "{{PROJECT_NAME}}",
val version: String = "1.0.0",
val description: String = "{{PROJECT_DESCRIPTION}}"
)
fun loadConfig(): Config {
return Config(
name = System.getenv("SERVER_NAME") ?: "{{PROJECT_NAME}}",
version = System.getenv("VERSION") ?: "1.0.0",
description = System.getenv("DESCRIPTION") ?: "{{PROJECT_DESCRIPTION}}"
)
}
package com.example.myserver.tools
import io.modelcontextprotocol.kotlin.sdk.server.Server
import io.modelcontextprotocol.kotlin.sdk.CallToolRequest
import io.modelcontextprotocol.kotlin.sdk.CallToolResult
import io.modelcontextprotocol.kotlin.sdk.TextContent
import kotlinx.serialization.json.buildJsonObject
import kotlinx.serialization.json.put
import kotlinx.serialization.json.putJsonObject
import kotlinx.serialization.json.putJsonArray
fun Server.registerTool1() {
addTool(
name = "tool1",
description = "Description of what tool1 does",
inputSchema = buildJsonObject {
put("type", "object")
putJsonObject("properties") {
putJsonObject("param1") {
put("type", "string")
put("description", "First parameter")
}
putJsonObject("param2") {
put("type", "integer")
put("description", "Optional second parameter")
}
}
putJsonArray("required") {
add("param1")
}
}
) { request: CallToolRequest ->
// Extract and validate parameters
val param1 = request.params.arguments["param1"] as? String
?: throw IllegalArgumentException("param1 is required")
val param2 = (request.params.arguments["param2"] as? Number)?.toInt() ?: 0
// Perform tool logic
val result = performTool1Logic(param1, param2)
CallToolResult(
content = listOf(
TextContent(text = result)
)
)
}
}
private fun performTool1Logic(param1: String, param2: Int): String {
// Implement tool logic here
return "Processed: $param1 with value $param2"
}
package com.example.myserver.tools
import io.modelcontextprotocol.kotlin.sdk.server.Server
fun Server.registerTools() {
registerTool1()
registerTool2()
// Register additional tools here
}
package com.example.myserver
import kotlinx.coroutines.test.runTest
import kotlin.test.Test
import kotlin.test.assertEquals
import kotlin.test.assertFalse
class ServerTest {
@Test
fun `test server creation`() = runTest {
val config = Config(
name = "test-server",
version = "1.0.0",
description = "Test server"
)
val server = createServer(config)
assertEquals("test-server", server.serverInfo.name)
assertEquals("1.0.0", server.serverInfo.version)
}
@Test
fun `test tool1 execution`() = runTest {
val config = Config()
val server = createServer(config)
// Test tool execution
// Note: You'll need to implement proper testing utilities
// for calling tools in the server
}
}
# {{PROJECT_NAME}}
A Model Context Protocol (MCP) server built with Kotlin.
## Description
{{PROJECT_DESCRIPTION}}
## Requirements
- Java 17 or higher
- Kotlin 2.1.0
## Installation
Build the project:
\`\`\`bash
./gradlew build
\`\`\`
## Usage
Run the server with stdio transport:
\`\`\`bash
./gradlew run
\`\`\`
Or build and run the jar:
\`\`\`bash
./gradlew installDist
./build/install/{{PROJECT_NAME}}/bin/{{PROJECT_NAME}}
\`\`\`
## Configuration
Configure via environment variables:
- `SERVER_NAME`: Server name (default: "{{PROJECT_NAME}}")
- `VERSION`: Server version (default: "1.0.0")
- `DESCRIPTION`: Server description
## Available Tools
### tool1
{{TOOL1_DESCRIPTION}}
**Input:**
- `param1` (string, required): First parameter
- `param2` (integer, optional): Second parameter
**Output:**
- Text result of the operation
## Development
Run tests:
\`\`\`bash
./gradlew test
\`\`\`
Build:
\`\`\`bash
./gradlew build
\`\`\`
Run with auto-reload (development):
\`\`\`bash
./gradlew run --continuous
\`\`\`
## Multiplatform
This project uses Kotlin Multiplatform and can target JVM, Wasm, and iOS.
See `build.gradle.kts` for platform configuration.
## License
MIT
When generating a Kotlin MCP server:
build.gradle.kts with all dependenciesbuildJsonObject for tool schemasval transport = StdioServerTransport()
server.connect(transport)
embeddedServer(Netty, port = 8080) {
mcp {
Server(/*...*/) { "Description" }
}
}.start(wait = true)
For multiplatform projects, add to build.gradle.kts:
kotlin {
jvm()
js(IR) { nodejs() }
wasmJs()
sourceSets {
commonMain.dependencies {
implementation("io.modelcontextprotocol:kotlin-sdk:0.7.2")
}
}
}
Dispatch fresh subagents per task with two-stage review (spec compliance, then code quality) in the current session.
ai image generation
ai video generation
agent tools
ai avatar video
Optimize content for discovery, citation, and visibility across AI search platforms and LLM assistants.
Official skill for integrating Firebase AI Logic (Gemini API) into web applications. Covers setup, multimodal inference, structured output, and security.
Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).
Smart router across 11+ image models for text-to-image and image-to-image generation via RunComfy CLI.
Route across 10+ video models (HappyHorse, Kling, Seedance, Veo, Wan, Hailuo, Dreamina) for text-to-video, image-to-video, and video extension.
Create talking-head and avatar videos by routing audio + image across five specialized models.
Generate vocal songs, instrumentals, jingles, and game loops via smart model routing across ElevenLabs and ACE Step.
A tool to turn screen demos into automated workflows.
Automate Your Business Processes Effortlessly with Microsoft Power Automate
Deploy autonomous AI Agents effortlessly in your browser.
Automate Your Business with Cheat Layer's Cutting-Edge AI
Get sh*t done faster with AI-powered writing and task automation.
Create AI workflows with drag-and-drop simplicity.
Automate Your Tasks with AutoGPT by aliabid94
Extend your apps, accelerate work