Update SDK to 0.0.64.rc1

build.gradle.kts

@@ -4,5 +4,5 @@ plugins {
 
 allprojects {
     group = "com.llama.llamastack"
-    version = "0.0.63"
+    version = "0.0.64.rc1"
 }

llama-stack-client-kotlin-client-local/src/main/kotlin/com/llama/llamastack/client/local/InferenceServiceLocalImpl.kt

@@ -27,9 +27,11 @@ constructor(
     private var modelName: String = ""
 
     private var sequenceLengthKey: String = "seq_len"
+    private var stopToken: String = ""
 
     override fun onResult(p0: String?) {
         if (PromptFormatLocal.getStopTokens(modelName).any { it == p0 }) {
+            stopToken = p0!!
             onResultComplete = true
             return
         }
@@ -55,7 +57,7 @@ constructor(
         params: InferenceChatCompletionParams,
         requestOptions: RequestOptions
     ): InferenceChatCompletionResponse {
-        resultMessage = ""
+        clearElements()
         val mModule = clientOptions.llamaModule
         modelName = params.modelId()
         val formattedPrompt =
@@ -79,7 +81,7 @@ constructor(
         onResultComplete = false
         onStatsComplete = false
 
-        return buildInferenceChatCompletionResponse(resultMessage, statsMetric)
+        return buildInferenceChatCompletionResponse(resultMessage, statsMetric, stopToken)
     }
 
     override fun chatCompletionStreaming(
@@ -109,4 +111,9 @@ constructor(
     ): EmbeddingsResponse {
         TODO("Not yet implemented")
     }
+
+    fun clearElements() {
+        resultMessage = ""
+        stopToken = ""
+    }
 }

llama-stack-client-kotlin-client-local/src/main/kotlin/com/llama/llamastack/client/local/LlamaStackClientClientLocalImpl.kt

@@ -20,6 +20,18 @@ constructor(
         TODO("Not yet implemented")
     }
 
+    override fun toolgroups(): ToolgroupService {
+        TODO("Not yet implemented")
+    }
+
+    override fun tools(): ToolService {
+        TODO("Not yet implemented")
+    }
+
+    override fun toolRuntime(): ToolRuntimeService {
+        TODO("Not yet implemented")
+    }
+
     override fun telemetry(): TelemetryService {
         TODO("Not yet implemented")
     }

llama-stack-client-kotlin-client-local/src/main/kotlin/com/llama/llamastack/client/local/util/PromptFormatLocal.kt

@@ -1,6 +1,5 @@
 package com.llama.llamastack.client.local.util
 
-import com.llama.llamastack.models.CompletionMessage
 import com.llama.llamastack.models.InferenceChatCompletionParams
 import com.llama.llamastack.models.SystemMessage
 import com.llama.llamastack.models.UserMessage
@@ -81,7 +80,9 @@ object PromptFormatLocal {
                 }
                 message.isCompletionMessage() -> {
                     // assistant message
-                    val completionMessage: CompletionMessage? = message.completionMessage()
+                    val completionMessage:
+                        InferenceChatCompletionParams.Message.CompletionMessage? =
+                        message.completionMessage()
                     val content: String? = completionMessage?.content()?.string()
                     if (content != null) {
                         format =

llama-stack-client-kotlin-client-local/src/main/kotlin/com/llama/llamastack/client/local/util/ResponseUtil.kt

@@ -1,28 +1,40 @@
 package com.llama.llamastack.client.local.util
 
 import com.llama.llamastack.core.JsonValue
-import com.llama.llamastack.models.CompletionMessage
 import com.llama.llamastack.models.InferenceChatCompletionResponse
 import com.llama.llamastack.models.InterleavedContent
 import com.llama.llamastack.models.ToolCall
 import java.util.UUID
 
 fun buildInferenceChatCompletionResponse(
     response: String,
-    stats: Float
+    stats: Float,
+    stopToken: String
 ): InferenceChatCompletionResponse {
     // check for prefix [ and suffix ] if so then tool call.
     // parse for "toolName", "additionalProperties"
-
     var completionMessage =
         if (response.startsWith("[") && response.endsWith("]")) {
             // custom tool call
-            CompletionMessage.builder()
+            InferenceChatCompletionResponse.ChatCompletionResponse.CompletionMessage.builder()
                 .toolCalls(createCustomToolCalls(response))
                 .content(InterleavedContent.ofString(""))
+                .role(
+                    InferenceChatCompletionResponse.ChatCompletionResponse.CompletionMessage.Role
+                        .ASSISTANT
+                )
+                .stopReason(mapStopTokenToReason(stopToken))
                 .build()
         } else {
-            CompletionMessage.builder().content(InterleavedContent.ofString(response)).build()
+            InferenceChatCompletionResponse.ChatCompletionResponse.CompletionMessage.builder()
+                .toolCalls(listOf())
+                .content(InterleavedContent.ofString(response))
+                .role(
+                    InferenceChatCompletionResponse.ChatCompletionResponse.CompletionMessage.Role
+                        .ASSISTANT
+                )
+                .stopReason(mapStopTokenToReason(stopToken))
+                .build()
         }
 
     var inferenceChatCompletionResponse =
@@ -67,3 +79,18 @@ fun createCustomToolCalls(response: String): List<ToolCall> {
 
     return toolCalls.toList()
 }
+
+fun mapStopTokenToReason(
+    stopToken: String
+): InferenceChatCompletionResponse.ChatCompletionResponse.CompletionMessage.StopReason =
+    when (stopToken) {
+        "<|eot_id|>" ->
+            InferenceChatCompletionResponse.ChatCompletionResponse.CompletionMessage.StopReason
+                .END_OF_TURN
+        "<|eom_id|>" ->
+            InferenceChatCompletionResponse.ChatCompletionResponse.CompletionMessage.StopReason
+                .END_OF_MESSAGE
+        else ->
+            InferenceChatCompletionResponse.ChatCompletionResponse.CompletionMessage.StopReason
+                .OUT_OF_TOKENS
+    }

llama-stack-client-kotlin-core/src/main/kotlin/com/llama/llamastack/client/LlamaStackClientClient.kt

@@ -22,11 +22,20 @@ import com.llama.llamastack.services.blocking.ScoringService
 import com.llama.llamastack.services.blocking.ShieldService
 import com.llama.llamastack.services.blocking.SyntheticDataGenerationService
 import com.llama.llamastack.services.blocking.TelemetryService
+import com.llama.llamastack.services.blocking.ToolRuntimeService
+import com.llama.llamastack.services.blocking.ToolService
+import com.llama.llamastack.services.blocking.ToolgroupService
 
 interface LlamaStackClientClient {
 
     fun async(): LlamaStackClientClientAsync
 
+    fun toolgroups(): ToolgroupService
+
+    fun tools(): ToolService
+
+    fun toolRuntime(): ToolRuntimeService
+
     fun agents(): AgentService
 
     fun batchInference(): BatchInferenceService

llama-stack-client-kotlin-core/src/main/kotlin/com/llama/llamastack/client/LlamaStackClientClientAsync.kt

@@ -22,11 +22,20 @@ import com.llama.llamastack.services.async.ScoringServiceAsync
 import com.llama.llamastack.services.async.ShieldServiceAsync
 import com.llama.llamastack.services.async.SyntheticDataGenerationServiceAsync
 import com.llama.llamastack.services.async.TelemetryServiceAsync
+import com.llama.llamastack.services.async.ToolRuntimeServiceAsync
+import com.llama.llamastack.services.async.ToolServiceAsync
+import com.llama.llamastack.services.async.ToolgroupServiceAsync
 
 interface LlamaStackClientClientAsync {
 
     fun sync(): LlamaStackClientClient
 
+    fun toolgroups(): ToolgroupServiceAsync
+
+    fun tools(): ToolServiceAsync
+
+    fun toolRuntime(): ToolRuntimeServiceAsync
+
     fun agents(): AgentServiceAsync
 
     fun batchInference(): BatchInferenceServiceAsync

llama-stack-client-kotlin-core/src/main/kotlin/com/llama/llamastack/client/LlamaStackClientClientAsyncImpl.kt

@@ -44,6 +44,12 @@ import com.llama.llamastack.services.async.SyntheticDataGenerationServiceAsync
 import com.llama.llamastack.services.async.SyntheticDataGenerationServiceAsyncImpl
 import com.llama.llamastack.services.async.TelemetryServiceAsync
 import com.llama.llamastack.services.async.TelemetryServiceAsyncImpl
+import com.llama.llamastack.services.async.ToolRuntimeServiceAsync
+import com.llama.llamastack.services.async.ToolRuntimeServiceAsyncImpl
+import com.llama.llamastack.services.async.ToolServiceAsync
+import com.llama.llamastack.services.async.ToolServiceAsyncImpl
+import com.llama.llamastack.services.async.ToolgroupServiceAsync
+import com.llama.llamastack.services.async.ToolgroupServiceAsyncImpl
 
 class LlamaStackClientClientAsyncImpl
 constructor(
@@ -61,6 +67,16 @@ constructor(
     // Pass the original clientOptions so that this client sets its own User-Agent.
     private val sync: LlamaStackClientClient by lazy { LlamaStackClientClientImpl(clientOptions) }
 
+    private val toolgroups: ToolgroupServiceAsync by lazy {
+        ToolgroupServiceAsyncImpl(clientOptionsWithUserAgent)
+    }
+
+    private val tools: ToolServiceAsync by lazy { ToolServiceAsyncImpl(clientOptionsWithUserAgent) }
+
+    private val toolRuntime: ToolRuntimeServiceAsync by lazy {
+        ToolRuntimeServiceAsyncImpl(clientOptionsWithUserAgent)
+    }
+
     private val agents: AgentServiceAsync by lazy {
         AgentServiceAsyncImpl(clientOptionsWithUserAgent)
     }
@@ -141,6 +157,12 @@ constructor(
 
     override fun sync(): LlamaStackClientClient = sync
 
+    override fun toolgroups(): ToolgroupServiceAsync = toolgroups
+
+    override fun tools(): ToolServiceAsync = tools
+
+    override fun toolRuntime(): ToolRuntimeServiceAsync = toolRuntime
+
     override fun agents(): AgentServiceAsync = agents
 
     override fun batchInference(): BatchInferenceServiceAsync = batchInference

llama-stack-client-kotlin-core/src/main/kotlin/com/llama/llamastack/client/LlamaStackClientClientImpl.kt

@@ -44,6 +44,12 @@ import com.llama.llamastack.services.blocking.SyntheticDataGenerationService
 import com.llama.llamastack.services.blocking.SyntheticDataGenerationServiceImpl
 import com.llama.llamastack.services.blocking.TelemetryService
 import com.llama.llamastack.services.blocking.TelemetryServiceImpl
+import com.llama.llamastack.services.blocking.ToolRuntimeService
+import com.llama.llamastack.services.blocking.ToolRuntimeServiceImpl
+import com.llama.llamastack.services.blocking.ToolService
+import com.llama.llamastack.services.blocking.ToolServiceImpl
+import com.llama.llamastack.services.blocking.ToolgroupService
+import com.llama.llamastack.services.blocking.ToolgroupServiceImpl
 
 class LlamaStackClientClientImpl
 constructor(
@@ -63,6 +69,16 @@ constructor(
         LlamaStackClientClientAsyncImpl(clientOptions)
     }
 
+    private val toolgroups: ToolgroupService by lazy {
+        ToolgroupServiceImpl(clientOptionsWithUserAgent)
+    }
+
+    private val tools: ToolService by lazy { ToolServiceImpl(clientOptionsWithUserAgent) }
+
+    private val toolRuntime: ToolRuntimeService by lazy {
+        ToolRuntimeServiceImpl(clientOptionsWithUserAgent)
+    }
+
     private val agents: AgentService by lazy { AgentServiceImpl(clientOptionsWithUserAgent) }
 
     private val batchInference: BatchInferenceService by lazy {
@@ -125,6 +141,12 @@ constructor(
 
     override fun async(): LlamaStackClientClientAsync = async
 
+    override fun toolgroups(): ToolgroupService = toolgroups
+
+    override fun tools(): ToolService = tools
+
+    override fun toolRuntime(): ToolRuntimeService = toolRuntime
+
     override fun agents(): AgentService = agents
 
     override fun batchInference(): BatchInferenceService = batchInference

llama-stack-client-kotlin-core/src/main/kotlin/com/llama/llamastack/core/ClientOptions.kt

@@ -64,6 +64,12 @@ private constructor(
 
         fun baseUrl(baseUrl: String) = apply { this.baseUrl = baseUrl }
 
+        fun responseValidation(responseValidation: Boolean) = apply {
+            this.responseValidation = responseValidation
+        }
+
+        fun maxRetries(maxRetries: Int) = apply { this.maxRetries = maxRetries }
+
         fun headers(headers: Headers) = apply {
             this.headers.clear()
             putAllHeaders(headers)
@@ -144,12 +150,6 @@ private constructor(
 
         fun removeAllQueryParams(keys: Set<String>) = apply { queryParams.removeAll(keys) }
 
-        fun responseValidation(responseValidation: Boolean) = apply {
-            this.responseValidation = responseValidation
-        }
-
-        fun maxRetries(maxRetries: Int) = apply { this.maxRetries = maxRetries }
-
         fun fromEnv() = apply {}
 
         fun build(): ClientOptions {