feat(agent): add image generation tool

src/main/presenter/agentRepository/index.ts

@@ -72,6 +72,8 @@ const mergeDeepChatConfig = (
     defaultModelPreset: overrideConfig.defaultModelPreset ?? baseConfig.defaultModelPreset ?? null,
     assistantModel: overrideConfig.assistantModel ?? baseConfig.assistantModel ?? null,
     visionModel: overrideConfig.visionModel ?? baseConfig.visionModel ?? null,
+    imageGenerationModel:
+      overrideConfig.imageGenerationModel ?? baseConfig.imageGenerationModel ?? null,
     defaultProjectPath: overrideConfig.defaultProjectPath ?? baseConfig.defaultProjectPath ?? null,
     systemPrompt: overrideConfig.systemPrompt ?? baseConfig.systemPrompt ?? '',
     permissionMode: overrideConfig.permissionMode ?? baseConfig.permissionMode ?? 'full_access',

src/main/presenter/agentRuntimePresenter/dispatch.ts

@@ -27,6 +27,10 @@ import { buildTerminalErrorBlocks } from './messageStore'
 import { finalizeTrailingPendingNarrativeBlocks } from './accumulator'
 import type { EchoHandle } from './echo'
 import { cloneBlocksForRenderer } from './echo'
+import {
+  insertBlocksAfterToolCall,
+  prepareToolImagePreviewPresentation
+} from './imageGenerationBlocks'
 import {
   buildAssistantPreviewMarkdown,
   buildAssistantResponseMarkdown,
@@ -56,6 +60,8 @@ type ToolExecutionContext = {
 type StagedToolResult = {
   toolCallId: string
   toolName: string
+  toolSource?: 'mcp' | 'agent'
+  serverName?: string
   toolArgs: string
   responseText: string
   isError: boolean
@@ -431,6 +437,8 @@ function buildToolErrorOutcome(
     stagedResult: {
       toolCallId: execution.completedToolCall.id,
       toolName: execution.completedToolCall.name,
+      toolSource: execution.toolDef?.source,
+      serverName: execution.toolContext.serverName,
       toolArgs: execution.completedToolCall.arguments,
       responseText: `Error: ${errorText}`,
       isError: true,
@@ -535,6 +543,13 @@ function applyFinalizedToolResults(params: {
       }
     }
 
+    const imagePresentation = prepareToolImagePreviewPresentation({
+      toolName: stagedResult.toolName,
+      toolSource: stagedResult.toolSource,
+      serverName: stagedResult.serverName,
+      isError: fittedResult.isError,
+      imagePreviews: stagedResult.imagePreviews
+    })
     updateToolCallBlock(
       state.blocks,
       fittedResult.toolCallId,
@@ -547,7 +562,12 @@ function applyFinalizedToolResults(params: {
             rtkMode: stagedResult.rtkMode,
             rtkFallbackReason: stagedResult.rtkFallbackReason
           },
-      stagedResult.imagePreviews
+      imagePresentation.toolBlockImagePreviews
+    )
+    insertBlocksAfterToolCall(
+      state.blocks,
+      fittedResult.toolCallId,
+      imagePresentation.promotedBlocks
     )
 
     if (fittedResult.isError) {
@@ -890,6 +910,8 @@ async function runToolCall(params: {
       stagedResult: {
         toolCallId: completedToolCall.id,
         toolName: completedToolCall.name,
+        toolSource: execution.toolDef?.source,
+        serverName: toolContext.serverName,
         toolArgs: completedToolCall.arguments,
         responseText: stagedResponseText,
         isError: stagedIsError,
@@ -1112,12 +1134,8 @@ export async function executeTools(
   for (const tc of state.completedToolCalls) {
     if (io.abortSignal.aborted) break
 
-    const { toolCall, toolContext, completedToolCall } = buildToolExecutionContext(
-      tc,
-      tools,
-      io.sessionId,
-      providerId
-    )
+    const execution = buildToolExecutionContext(tc, tools, io.sessionId, providerId)
+    const { toolCall, toolContext } = execution
 
     try {
       if (toolCall.function.name === QUESTION_TOOL_NAME) {
@@ -1192,11 +1210,7 @@ export async function executeTools(
       })
 
       const outcome = await runToolCall({
-        execution: {
-          toolCall,
-          toolContext,
-          completedToolCall
-        },
+        execution,
         toolPresenter,
         permissionMode,
         hooks,

src/main/presenter/agentRuntimePresenter/imageGenerationBlocks.ts

@@ -0,0 +1,87 @@
+import type { AssistantMessageBlock } from '@shared/types/agent-interface'
+import type { ToolCallImagePreview } from '@shared/types/core/mcp'
+import {
+  IMAGE_GENERATE_TOOL_NAME,
+  IMAGE_GENERATION_TOOL_SERVER_NAME
+} from '@shared/agentImageGenerationTool'
+
+export function prepareToolImagePreviewPresentation(params: {
+  toolName: string
+  toolSource?: 'mcp' | 'agent'
+  serverName?: string
+  isError: boolean
+  imagePreviews?: ToolCallImagePreview[]
+}): {
+  toolBlockImagePreviews?: ToolCallImagePreview[]
+  promotedBlocks: AssistantMessageBlock[]
+} {
+  const { toolName, toolSource, serverName, isError, imagePreviews } = params
+  if (!imagePreviews) {
+    return { promotedBlocks: [] }
+  }
+
+  if (
+    toolName !== IMAGE_GENERATE_TOOL_NAME ||
+    toolSource !== 'agent' ||
+    serverName !== IMAGE_GENERATION_TOOL_SERVER_NAME ||
+    isError ||
+    imagePreviews.length === 0
+  ) {
+    return {
+      toolBlockImagePreviews: imagePreviews,
+      promotedBlocks: []
+    }
+  }
+
+  const timestamp = Date.now()
+  const promotedBlocks: AssistantMessageBlock[] = []
+  for (const preview of imagePreviews) {
+    const { data, mimeType } = preview
+    if (!data || !mimeType) {
+      continue
+    }
+
+    promotedBlocks.push({
+      type: 'image',
+      content: '',
+      status: 'success',
+      timestamp,
+      image_data: {
+        data,
+        mimeType
+      }
+    })
+  }
+
+  if (promotedBlocks.length === 0) {
+    return {
+      toolBlockImagePreviews: imagePreviews,
+      promotedBlocks: []
+    }
+  }
+
+  return {
+    toolBlockImagePreviews: [],
+    promotedBlocks
+  }
+}
+
+export function insertBlocksAfterToolCall(
+  blocks: AssistantMessageBlock[],
+  toolCallId: string,
+  newBlocks: AssistantMessageBlock[]
+): void {
+  if (newBlocks.length === 0) {
+    return
+  }
+
+  const toolBlockIndex = blocks.findIndex(
+    (block) => block.type === 'tool_call' && block.tool_call?.id === toolCallId
+  )
+  if (toolBlockIndex === -1) {
+    blocks.push(...newBlocks)
+    return
+  }
+
+  blocks.splice(toolBlockIndex + 1, 0, ...newBlocks)
+}

src/main/presenter/agentRuntimePresenter/index.ts

@@ -103,6 +103,10 @@ import {
   emitDeepChatInternalSessionUpdate,
   extractWaitingInteraction
 } from './internalSessionEvents'
+import {
+  insertBlocksAfterToolCall,
+  prepareToolImagePreviewPresentation
+} from './imageGenerationBlocks'
 
 type PendingInteractionEntry = {
   interaction: PendingToolInteraction
@@ -112,6 +116,8 @@ type PendingInteractionEntry = {
 type DeferredToolExecutionResult = {
   responseText: string
   isError: boolean
+  toolSource?: 'mcp' | 'agent'
+  serverName?: string
   offloadPath?: string
   rtkApplied?: boolean
   rtkMode?: 'rewrite' | 'direct' | 'bypass'
@@ -1029,6 +1035,14 @@ export class AgentRuntimePresenter implements IAgentImplementation {
             this.setSessionStatus(sessionId, 'error')
             return { resumed: false }
           }
+          const imagePresentation = prepareToolImagePreviewPresentation({
+            toolName: toolCall.name || '',
+            toolSource: execution.toolSource,
+            serverName: execution.serverName,
+            isError: execution.isError,
+            imagePreviews: execution.imagePreviews
+          })
+
           this.updateToolCallResponse(
             blocks,
             toolCall.id,
@@ -1038,9 +1052,10 @@ export class AgentRuntimePresenter implements IAgentImplementation {
               rtkApplied: execution.rtkApplied,
               rtkMode: execution.rtkMode,
               rtkFallbackReason: execution.rtkFallbackReason,
-              imagePreviews: execution.imagePreviews
+              imagePreviews: imagePresentation.toolBlockImagePreviews
             }
           )
+          insertBlocksAfterToolCall(blocks, toolCall.id, imagePresentation.promotedBlocks)
           resumeBudgetToolCall = {
             id: toolCall.id,
             name: toolCall.name || '',
@@ -4243,6 +4258,8 @@ export class AgentRuntimePresenter implements IAgentImplementation {
       return {
         responseText: prepared.content,
         isError: Boolean(rawData.isError),
+        toolSource: toolDefinition.source,
+        serverName: toolDefinition.server.name,
         offloadPath: prepared.offloadPath,
         rtkApplied: rawData.rtkApplied,
         rtkMode: rawData.rtkMode,

src/main/presenter/configPresenter/index.ts

@@ -892,6 +892,10 @@ export class ConfigPresenter implements IConfigPresenter {
       updates.visionModel = null
     }
 
+    if (isDeprecatedBuiltinModelSelection(config.imageGenerationModel)) {
+      updates.imageGenerationModel = null
+    }
+
     if (Object.keys(updates).length > 0) {
       this.updateBuiltinDeepChatConfig(updates)
     }

src/main/presenter/index.ts

@@ -380,6 +380,14 @@ export class Presenter implements IPresenter {
             temperature,
             maxTokens,
             options
+          ),
+        generateImageStandalone: (providerId, prompt, modelId, imageOptions, options) =>
+          this.llmproviderPresenter.generateImageStandalone(
+            providerId,
+            prompt,
+            modelId,
+            imageOptions,
+            options
           )
       }),
       cacheImage: (data) => this.devicePresenter.cacheImage(data),