Kilo-Org · kiloconnect · Dec 20, 2025
Changes from all commits

Commits

Show all changes
1 commit

Select commit

550f7ef
fix: remove duplicate tool processing from 7 additional providers
kiloconnect[bot] Dec 20, 2025
diff --git a/.changeset/fix-duplicate-tool-processing-all-providers.md b/.changeset/fix-duplicate-tool-processing-all-providers.md
@@ -0,0 +1,16 @@
+---
+"kilo-code": patch
+---
+
+Fix duplicate tool processing in 7 additional providers (inception, lm-studio, deepinfra, xai, lite-llm, qwen-code, chutes)
+
+Following PR #4531 which fixed duplicate tool processing in OpenAI-compatible providers, this change applies the same fix to 7 additional providers that had the same issue. The `ToolCallAccumulator` was processing tool calls in the streaming loop, but the code was also manually emitting `tool_call_partial` chunks, causing duplicates. This fix removes the `ToolCallAccumulator` usage and relies solely on emitting raw `tool_call_partial` chunks, which are then properly handled by `NativeToolCallParser`.
+
+Affected providers:
+- inception
+- lm-studio  
+- deepinfra
+- xai
+- lite-llm
+- qwen-code
+- chutes
diff --git a/src/api/providers/base-openai-compatible-provider.ts b/src/api/providers/base-openai-compatible-provider.ts
@@ -14,7 +14,6 @@ import { BaseProvider } from "./base-provider"
 import { verifyFinishReason } from "./kilocode/verifyFinishReason"
 import { handleOpenAIError } from "./utils/openai-error-handler"
 import { fetchWithTimeout } from "./kilocode/fetchWithTimeout" // kilocode_change
-import { ToolCallAccumulator } from "./kilocode/nativeToolCallHelpers" // kilocode_change
 import { calculateApiCostOpenAI } from "../../shared/cost"
 import { getApiRequestTimeout } from "./utils/timeout-config"
 

diff --git a/src/api/providers/chutes.ts b/src/api/providers/chutes.ts
@@ -11,7 +11,7 @@ import { ApiStream } from "../transform/stream"
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
 
 import { RouterProvider } from "./router-provider"
-import { addNativeToolCallsToParams, ToolCallAccumulator } from "./kilocode/nativeToolCallHelpers"
+import { addNativeToolCallsToParams } from "./kilocode/nativeToolCallHelpers"
 
 export class ChutesHandler extends RouterProvider implements SingleCompletionHandler {
 	constructor(options: ApiHandlerOptions) {
@@ -69,8 +69,6 @@ export class ChutesHandler extends RouterProvider implements SingleCompletionHan
 	): ApiStream {
 		const model = await this.fetchModel()
 
-		const toolCallAccumulator = new ToolCallAccumulator() // kilocode_change
-
 		if (model.id.includes("DeepSeek-R1")) {
 			const stream = await this.client.chat.completions.create({
 				...this.getCompletionParams(systemPrompt, messages, metadata),
@@ -89,8 +87,6 @@ export class ChutesHandler extends RouterProvider implements SingleCompletionHan
 			for await (const chunk of stream) {
 				const delta = chunk.choices[0]?.delta
 
-				yield* toolCallAccumulator.processChunk(chunk) // kilocode_change
-
 				if (delta?.content) {
 					for (const processedChunk of matcher.update(delta.content)) {
 						yield processedChunk
@@ -132,8 +128,6 @@ export class ChutesHandler extends RouterProvider implements SingleCompletionHan
 			for await (const chunk of stream) {
 				const delta = chunk.choices[0]?.delta
 
-				yield* toolCallAccumulator.processChunk(chunk) // kilocode_change
-
 				if (delta?.content) {
 					yield { type: "text", text: delta.content }
 				}

diff --git a/src/api/providers/deepinfra.ts b/src/api/providers/deepinfra.ts
@@ -13,7 +13,7 @@ import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from ".
 import { RouterProvider } from "./router-provider"
 import { getModelParams } from "../transform/model-params"
 import { getModels } from "./fetchers/modelCache"
-import { addNativeToolCallsToParams, ToolCallAccumulator } from "./kilocode/nativeToolCallHelpers"
+import { addNativeToolCallsToParams } from "./kilocode/nativeToolCallHelpers"
 
 export class DeepInfraHandler extends RouterProvider implements SingleCompletionHandler {
 	constructor(options: ApiHandlerOptions) {
@@ -94,12 +94,9 @@ export class DeepInfraHandler extends RouterProvider implements SingleCompletion
 		const { data: stream } = await this.client.chat.completions.create(requestOptions).withResponse()
 
 		let lastUsage: OpenAI.CompletionUsage | undefined
-		const toolCallAccumulator = new ToolCallAccumulator() // kilocode_change
 		for await (const chunk of stream) {
 			const delta = chunk.choices[0]?.delta
 
-			yield* toolCallAccumulator.processChunk(chunk) // kilocode_change
-
 			if (delta?.content) {
 				yield { type: "text", text: delta.content }
 			}
@@ -108,7 +105,7 @@ export class DeepInfraHandler extends RouterProvider implements SingleCompletion
 				yield { type: "reasoning", text: (delta.reasoning_content as string | undefined) || "" }
 			}
 
-			// Handle tool calls in stream - emit partial chunks for NativeToolCallParser
+			// Emit raw tool call chunks - NativeToolCallParser handles state management
 			if (delta?.tool_calls) {
 				for (const toolCall of delta.tool_calls) {
 					yield {

diff --git a/src/api/providers/inception.ts b/src/api/providers/inception.ts
@@ -13,7 +13,7 @@ import Anthropic from "@anthropic-ai/sdk"
 import { ApiStream, ApiStreamUsageChunk } from "../transform/stream"
 import OpenAI from "openai"
 import { convertToOpenAiMessages } from "../transform/openai-format"
-import { addNativeToolCallsToParams, ToolCallAccumulator } from "./kilocode/nativeToolCallHelpers"
+import { addNativeToolCallsToParams } from "./kilocode/nativeToolCallHelpers"
 
 export class InceptionLabsHandler extends RouterProvider implements SingleCompletionHandler {
 	constructor(options: ApiHandlerOptions) {
@@ -83,10 +83,8 @@ export class InceptionLabsHandler extends RouterProvider implements SingleComple
 			.withResponse()
 
 		let lastUsage: OpenAI.CompletionUsage | undefined
-		const toolCallAccumulator = new ToolCallAccumulator()
 		for await (const chunk of stream) {
 			const delta = chunk.choices[0]?.delta
-			yield* toolCallAccumulator.processChunk(chunk)
 
 			if (delta?.content) {
 				yield { type: "text", text: delta.content }
@@ -95,6 +93,20 @@ export class InceptionLabsHandler extends RouterProvider implements SingleComple
 			if (delta && "reasoning_content" in delta && delta.reasoning_content) {
 				yield { type: "reasoning", text: (delta.reasoning_content as string | undefined) || "" }
 			}
+
+			// Emit raw tool call chunks - NativeToolCallParser handles state management
+			if (delta?.tool_calls) {
+				for (const toolCall of delta.tool_calls) {
+					yield {
+						type: "tool_call_partial",
+						index: toolCall.index,
+						id: toolCall.id,
+						name: toolCall.function?.name,
+						arguments: toolCall.function?.arguments,
+					}
+				}
+			}
+
 			if (chunk.usage) {
 				lastUsage = chunk.usage
 			}

diff --git a/src/api/providers/lite-llm.ts b/src/api/providers/lite-llm.ts
@@ -12,7 +12,7 @@ import { convertToOpenAiMessages } from "../transform/openai-format"
 
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
 import { RouterProvider } from "./router-provider"
-import { addNativeToolCallsToParams, ToolCallAccumulator } from "./kilocode/nativeToolCallHelpers"
+import { addNativeToolCallsToParams } from "./kilocode/nativeToolCallHelpers"
 
 /**
  * LiteLLM provider handler
@@ -153,18 +153,15 @@ export class LiteLLMHandler extends RouterProvider implements SingleCompletionHa
 
 			let lastUsage
 
-			const toolCallAccumulator = new ToolCallAccumulator() // kilocode_change
 			for await (const chunk of completion) {
 				const delta = chunk.choices[0]?.delta
 				const usage = chunk.usage as LiteLLMUsage
 
-				yield* toolCallAccumulator.processChunk(chunk) // kilocode_change
-
 				if (delta?.content) {
 					yield { type: "text", text: delta.content }
 				}
 
-				// Handle tool calls in stream - emit partial chunks for NativeToolCallParser
+				// Emit raw tool call chunks - NativeToolCallParser handles state management
 				if (delta?.tool_calls) {
 					for (const toolCall of delta.tool_calls) {
 						yield {

diff --git a/src/api/providers/lm-studio.ts b/src/api/providers/lm-studio.ts
@@ -14,7 +14,7 @@ import { ApiStream } from "../transform/stream"
 import { BaseProvider } from "./base-provider"
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
 import { fetchWithTimeout, HeadersTimeoutError } from "./kilocode/fetchWithTimeout"
-import { addNativeToolCallsToParams, ToolCallAccumulator } from "./kilocode/nativeToolCallHelpers"
+import { addNativeToolCallsToParams } from "./kilocode/nativeToolCallHelpers"
 import { getModels, getModelsFromCache } from "./fetchers/modelCache"
 import { handleOpenAIError } from "./utils/openai-error-handler"
 import { getApiRequestTimeout } from "./utils/timeout-config" // kilocode_change
@@ -114,18 +114,28 @@ export class LmStudioHandler extends BaseProvider implements SingleCompletionHan
 					}) as const,
 			)
 
-			const toolCallAccumulator = new ToolCallAccumulator() // kilocode_change
 			for await (const chunk of results) {
 				const delta = chunk.choices[0]?.delta
 
-				yield* toolCallAccumulator.processChunk(chunk) // kilocode_change
-
 				if (delta?.content) {
 					assistantText += delta.content
 					for (const processedChunk of matcher.update(delta.content)) {
 						yield processedChunk
 					}
 				}
+
+				// Emit raw tool call chunks - NativeToolCallParser handles state management
+				if (delta?.tool_calls) {
+					for (const toolCall of delta.tool_calls) {
+						yield {
+							type: "tool_call_partial",
+							index: toolCall.index,
+							id: toolCall.id,
+							name: toolCall.function?.name,
+							arguments: toolCall.function?.arguments,
+						}
+					}
+				}
 			}
 
 			for (const processedChunk of matcher.final()) {

diff --git a/src/api/providers/qwen-code.ts b/src/api/providers/qwen-code.ts
@@ -16,7 +16,7 @@ import type {
 	ApiHandlerCreateMessageMetadata, // kilocode_change
 	SingleCompletionHandler,
 } from "../index"
-import { addNativeToolCallsToParams, ToolCallAccumulator } from "./kilocode/nativeToolCallHelpers"
+import { addNativeToolCallsToParams } from "./kilocode/nativeToolCallHelpers"
 
 const QWEN_OAUTH_BASE_URL = "https://chat.qwen.ai"
 const QWEN_OAUTH_TOKEN_ENDPOINT = `${QWEN_OAUTH_BASE_URL}/api/v1/oauth2/token`
@@ -235,7 +235,6 @@ export class QwenCodeHandler extends BaseProvider implements SingleCompletionHan
 		const stream = await this.callApiWithRetry(() => client.chat.completions.create(requestOptions))
 
 		let fullContent = ""
-		const toolCallAccumulator = new ToolCallAccumulator() // kilocode_change
 
 		for await (const apiChunk of stream) {
 			const delta = apiChunk.choices[0]?.delta ?? {}
@@ -285,7 +284,18 @@ export class QwenCodeHandler extends BaseProvider implements SingleCompletionHan
 				}
 			}
 
-			yield* toolCallAccumulator.processChunk(apiChunk) // kilocode_change
+			// Emit raw tool call chunks - NativeToolCallParser handles state management
+			if (delta.tool_calls) {
+				for (const toolCall of delta.tool_calls) {
+					yield {
+						type: "tool_call_partial",
+						index: toolCall.index,
+						id: toolCall.id,
+						name: toolCall.function?.name,
+						arguments: toolCall.function?.arguments,
+					}
+				}
+			}
 
 			if (apiChunk.usage) {
 				yield {

diff --git a/src/api/providers/xai.ts b/src/api/providers/xai.ts
@@ -15,7 +15,7 @@ import { BaseProvider } from "./base-provider"
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
 import { verifyFinishReason } from "./kilocode/verifyFinishReason" // kilocode_change
 import { handleOpenAIError } from "./utils/openai-error-handler"
-import { addNativeToolCallsToParams, ToolCallAccumulator } from "./kilocode/nativeToolCallHelpers"
+import { addNativeToolCallsToParams } from "./kilocode/nativeToolCallHelpers"
 
 const XAI_DEFAULT_TEMPERATURE = 0
 
@@ -84,14 +84,11 @@ export class XAIHandler extends BaseProvider implements SingleCompletionHandler
 			throw handleOpenAIError(error, this.providerName)
 		}
 
-		const toolCallAccumulator = new ToolCallAccumulator() // kilocode_change
 		for await (const chunk of stream) {
 			verifyFinishReason(chunk.choices[0]) // kilocode_change
 			const delta = chunk.choices[0]?.delta
 			const finishReason = chunk.choices[0]?.finish_reason
 
-			yield* toolCallAccumulator.processChunk(chunk) // kilocode_change
-
 			if (delta?.content) {
 				yield {
 					type: "text",
@@ -106,7 +103,7 @@ export class XAIHandler extends BaseProvider implements SingleCompletionHandler
 				}
 			}
 
-			// Handle tool calls in stream - emit partial chunks for NativeToolCallParser
+			// Emit raw tool call chunks - NativeToolCallParser handles state management
 			if (delta?.tool_calls) {
 				for (const toolCall of delta.tool_calls) {
 					yield {