Agent Chat

\n {isLoading && (\n \n Cancel\n \n )}\n

\n\n {error && (\n \n

Error

{error}

\n setError(null)}\n className=\"ml-4 text-red-500 hover:text-red-700 font-bold text-xl self-start\"\n aria-label=\"Dismiss error\"\n >\n ×\n \n

\n )}\n\n {/* Message Display Area */}\n

\n {messages.map((message, index) => (\n \n {/* Improved rendering for tool results */}\n {message.role === \"tool\" && typeof message.content === \"string\" && message.content.includes(\"```json\")\n ? (\n <>\n {/* Text before JSON block */}\n {message.content.substring(0, message.content.indexOf(\"```json\"))}\n {/* JSON block */}\n

\n                        {message.content.substring(message.content.indexOf(\"```json\") + 7, message.content.lastIndexOf(\"```\")).trim()}\n

\n \n )\n : message.role === \"tool\" // Render non-JSON tool messages or errors\n ?

{message.content}

\n : ( // Render user, agent, system messages\n // Handle potential line breaks in agent/user messages\n message.content.split(\"\\n\").map((line, i) => (\n {line}{i === message.content.split(\"\\n\").length - 1 ? \"\" :
}\n )) || (message.isStreaming ? \"...\" : \"\") // Show ellipsis if streaming content is empty\n )}\n

\n ))}\n {/* Dummy div to help with scrolling to the bottom */}\n

\n \n\n {/* Input Area */}\n

\n setInputMessage(e.target.value)}\n onKeyPress={(e) => e.key === \"Enter\" && !isLoading && sendMessage()}\n placeholder=\"Ask me anything...\"\n disabled={isLoading}\n aria-label=\"Message input\"\n />\n \n {isLoading ? \"Processing...\" : \"Send\"}\n \n

\n \n );\n}\n\n// --- Client-side initialization ---\nfunction client() {\n const rootElement = document.getElementById(\"root\");\n if (rootElement) {\n createRoot(rootElement).render();\n } else {\n console.error(\"Root element not found for React app\");\n // Optionally create a root element if it doesn't exist\n const newRoot = document.createElement(\"div\");\n newRoot.id = \"root\";\n document.body.appendChild(newRoot);\n createRoot(newRoot).render();\n log(\"WARN\", \"Created missing #root element for React app.\");\n }\n}\n// Ensure client code runs only in browser environment\nif (typeof document !== \"undefined\" && document.getElementById(\"root\")) {\n client();\n}\n\n// ========================================================================\n// SERVER CODE (Val Town Endpoint) - CORRECTED & CONSOLIDATED\n// ========================================================================\nexport default async function server(request: Request): Promise {\n // --- Server-side Logging ---\n function serverLog(level: \"INFO\" | \"WARN\" | \"ERROR\", message: string, context?: any) {\n const timestamp = new Date().toISOString();\n let contextString: string | undefined;\n try {\n // Avoid overly large contexts in logs\n const contextStrFull = context ? JSON.stringify(context) : undefined;\n contextString = contextStrFull\n ? contextStrFull.substring(0, 500) + (contextStrFull.length > 500 ? \"...\" : \"\")\n : undefined;\n } catch (e) {\n contextString = \"[Context could not be stringified]\";\n }\n const logEntry = { timestamp, level, message, context: contextString };\n console.log(JSON.stringify(logEntry));\n return logEntry;\n }\n\n const openai = new OpenAI();\n\n // --- Centralized OpenAI Call Function ---\n async function callOpenAI(\n systemPrompt: string,\n userMessage: string,\n model = \"gpt-4o-mini\",\n expectJson = false,\n isStreaming = false,\n ): Promise<{ role: \"assistant\" | \"system\"; content: string } | ReadableStream> {\n try {\n serverLog(\"INFO\", \"Calling OpenAI\", {\n model,\n promptStart: systemPrompt.substring(0, 50).replace(/\\s+/g, \" \") + \"...\",\n expectJson,\n isStreaming,\n });\n\n const response = await openai.chat.completions.create({\n model: model,\n response_format: expectJson && !isStreaming ? { type: \"json_object\" } : undefined,\n messages: [\n { role: \"system\", content: systemPrompt },\n { role: \"user\", content: userMessage },\n ],\n temperature: 0.3,\n stream: isStreaming,\n });\n\n if (isStreaming) {\n serverLog(\"INFO\", \"OpenAI call initiated (Streaming)\");\n if (response instanceof ReadableStream) {\n return response;\n } else if (Symbol.asyncIterator in response) {\n const iterator = response[Symbol.asyncIterator]();\n return new ReadableStream({\n async pull(controller) {\n try {\n const { done, value } = await iterator.next();\n if (done) {\n controller.close();\n serverLog(\"INFO\", \"OpenAI Stream finished.\");\n } else {\n const content = value?.choices?.[0]?.delta?.content || \"\";\n if (content) {\n controller.enqueue(new TextEncoder().encode(content));\n }\n }\n } catch (err) {\n serverLog(\"ERROR\", \"Error reading from OpenAI stream iterator\", { error: (err as Error).message });\n controller.error(err);\n }\n },\n });\n } else {\n throw new Error(\"OpenAI response is not a readable stream or async iterator when streaming was expected.\");\n }\n } else {\n const content = (response as any).choices?.[0]?.message?.content;\n if (!content) {\n serverLog(\"ERROR\", \"OpenAI API returned empty content (non-streaming)\", { response });\n throw new Error(\"Received invalid or empty response from AI model.\");\n }\n\n if (expectJson) {\n try {\n JSON.parse(content);\n serverLog(\"INFO\", \"OpenAI call successful (JSON validated)\");\n return { role: \"assistant\", content };\n } catch (jsonError) {\n serverLog(\"ERROR\", \"OpenAI response is not valid JSON\", {\n contentPreview: content.substring(0, 200) + \"...\",\n error: (jsonError as Error).message,\n });\n // Using concat\n throw new Error(\n \"AI model did not return valid JSON. Content preview: \" + content.substring(0, 100) + \"...\",\n );\n }\n } else {\n serverLog(\"INFO\", \"OpenAI call successful (Text)\");\n return { role: \"assistant\", content };\n }\n }\n } catch (error) {\n const errorMessageString = error instanceof Error ? error.message : String(error);\n const errorStatus = error instanceof Error ? (error as any).status ?? (error as any).response?.status : undefined;\n serverLog(\"ERROR\", \"OpenAI API call failed\", {\n model,\n error: errorMessageString,\n status: errorStatus,\n isStreaming,\n });\n\n // Using concat for user-friendly messages\n let userFriendlyErrorMessage = \"Error with AI model (\" + model + \").\";\n const statusCode = errorStatus;\n\n if (statusCode === 401) userFriendlyErrorMessage = \"OpenAI Auth Error (401). Check Val Town 'openai' secret.\";\n else if (statusCode === 429)\n userFriendlyErrorMessage = \"OpenAI Rate Limit/Quota Error (429). Check OpenAI plan/usage or try again later.\";\n else if (statusCode === 400)\n userFriendlyErrorMessage = \"OpenAI Bad Request (400). Possible issue with request or content filter. Details: \"\n + errorMessageString;\n else if (errorMessageString?.includes(\"maximum context length\"))\n userFriendlyErrorMessage = \"OpenAI Context Length Error (400). The request is too long.\";\n else if (statusCode >= 500)\n userFriendlyErrorMessage = \"OpenAI Server Error (\" + statusCode\n + \"). OpenAI might be having issues. Try again later.\";\n else if (error instanceof Error && (error as any).code === \"ENOTFOUND\")\n userFriendlyErrorMessage = \"Network Error (ENOTFOUND). Cannot resolve OpenAI API address.\";\n else if (error instanceof Error && (error as any).code === \"ECONNREFUSED\")\n userFriendlyErrorMessage = \"Network Error (ECONNREFUSED). Cannot connect to OpenAI API.\";\n else if (errorMessageString?.includes(\"secret\")) userFriendlyErrorMessage = errorMessageString;\n else userFriendlyErrorMessage = userFriendlyErrorMessage + \" Details: \" + errorMessageString;\n\n if (isStreaming) {\n throw new Error(userFriendlyErrorMessage);\n } else {\n // Using concat\n return { role: \"system\", content: \"System Error: \" + userFriendlyErrorMessage };\n }\n }\n }\n\n // --- Agent Prompts (Refined) ---\n const schemaDefinerPrompt = `\nYou are SchemaDefiner, an expert data modeler. Create a detailed JSON schema for the requested topic.\nFocus on common, relevant fields with types (string, number, boolean, array, object), descriptions, and nested structures ('items' for arrays, 'properties' for objects).\nTopic: {{TOPIC}}\nOutput ONLY the raw JSON schema object, starting with { and ending with }.`;\n\n const dataFillerPrompt = `\nYou are DataFiller. Populate the JSON schema with realistic, comprehensive data for the topic.\nGenerate plausible values for all fields. For arrays, provide a good number of diverse examples (e.g., 3-7 items unless context dictates otherwise). Be specific if the topic is narrow.\nTopic: {{TOPIC}}\nJSON Schema:\n\\`\\`\\`json\n{{SCHEMA}}\n\\`\\`\\`\nOutput ONLY the raw populated JSON object, starting with { and ending with }.`;\n\n const schemaValidatorPrompt = `\nYou are SchemaValidator. Verify the data object strictly adheres to the schema (structure, fields, types).\nAssume all schema fields are required.\nJSON Schema:\n\\`\\`\\`json\n{{SCHEMA}}\n\\`\\`\\`\nData Object:\n\\`\\`\\`json\n{{DATA}}\n\\`\\`\\`\nInstructions:\n1. Check for missing fields, extra fields, and type mismatches recursively.\n2. If valid: Output the original data object.\n3. If ONLY extra fields exist: Remove them and output the corrected JSON object.\n4. If missing fields or type mismatches: Output {\"error\": \"Validation failed: [Concise reason, e.g., Missing field 'xyz' or Field 'abc' expected number, got string.\"]}.\nOutput ONLY the resulting raw JSON object.`;\n\n const finalizerPrompt = `\nYou are Finalizer. Perform a final quality check on the validated data object for the given topic and URL path.\nOriginal Topic: {{TOPIC}}\nRequested URL Path: /{{PATH}}\nValidated Data Object:\n\\`\\`\\`json\n{{DATA}}\n\\`\\`\\`\nInstructions:\n1. Review data for coherence, plausibility, and completeness related to the topic/path.\n2. Ensure logical structure and clarity. Reflect path hierarchy if applicable.\n3. Correct minor inconsistencies, typos, or formatting. Remove placeholders.\n4. Make minor structural adjustments ONLY for clarity, respecting schema intent.\n5. Output ONLY the final, polished, raw JSON object, starting with { and ending with }.`;\n\n // --- Multi-Agent Workflow ---\n async function runDataGenerationWorkflow(\n topic: string,\n path: string,\n updateCallback?: (\n statusUpdate: { status: string; agent?: string; error?: boolean; message?: string },\n ) => Promise,\n ): Promise {\n serverLog(\"INFO\", \"Starting Data Generation Workflow\", { topic, path });\n const workflowState = { topic, path, schema: null, filledData: null, validatedData: null, finalData: null };\n\n const agents = [\n { name: \"SchemaDefiner\", prompt: schemaDefinerPrompt, outputKey: \"schema\", expectJson: true },\n { name: \"DataFiller\", prompt: dataFillerPrompt, outputKey: \"filledData\", expectJson: true },\n { name: \"SchemaValidator\", prompt: schemaValidatorPrompt, outputKey: \"validatedData\", expectJson: true },\n { name: \"Finalizer\", prompt: finalizerPrompt, outputKey: \"finalData\", expectJson: true },\n ];\n\n try {\n for (const agent of agents) {\n if (updateCallback) await updateCallback({ status: \"Running\", agent: agent.name });\n // Using concat\n serverLog(\"INFO\", \"Workflow Step: Running \" + agent.name);\n\n let systemPrompt = agent.prompt.replace(\"{{TOPIC}}\", workflowState.topic).replace(\n \"{{PATH}}\",\n workflowState.path,\n );\n // Using concat\n let userMessageContent = \"Process request for topic: \" + workflowState.topic;\n\n if (agent.name === \"DataFiller\") {\n if (!workflowState.schema) throw new Error(\"Workflow Error: Schema missing for DataFiller.\");\n systemPrompt = systemPrompt.replace(\"{{SCHEMA}}\", JSON.stringify(workflowState.schema, null, 2));\n // Using concat\n userMessageContent = \"Generate data for topic: \" + workflowState.topic\n + \" using the schema. Make it comprehensive.\";\n } else if (agent.name === \"SchemaValidator\") {\n if (!workflowState.schema) throw new Error(\"Workflow Error: Schema missing for SchemaValidator.\");\n if (!workflowState.filledData) throw new Error(\"Workflow Error: Filled data missing for SchemaValidator.\");\n systemPrompt = systemPrompt.replace(\"{{SCHEMA}}\", JSON.stringify(workflowState.schema, null, 2));\n systemPrompt = systemPrompt.replace(\"{{DATA}}\", JSON.stringify(workflowState.filledData, null, 2));\n userMessageContent = \"Validate the provided data against the schema.\";\n } else if (agent.name === \"Finalizer\") {\n if (!workflowState.validatedData) throw new Error(\"Workflow Error: Validated data missing for Finalizer.\");\n if (workflowState.validatedData.error) {\n // Using concat\n throw new Error(\"Schema validation failed prior to Finalizer: \" + workflowState.validatedData.error);\n }\n systemPrompt = systemPrompt.replace(\"{{DATA}}\", JSON.stringify(workflowState.validatedData, null, 2));\n // Using concat\n userMessageContent = \"Finalize the data for topic: \" + workflowState.topic + \" based on path: /\"\n + workflowState.path;\n }\n\n const response = await callOpenAI(systemPrompt, userMessageContent, \"gpt-4o-mini\", agent.expectJson, false);\n\n if (typeof response === \"string\" || response instanceof ReadableStream) {\n // Using concat\n throw new Error(\"Agent \" + agent.name + \" received unexpected stream response.\");\n }\n if (response.role === \"system\") {\n // Using concat\n throw new Error(\"Agent \" + agent.name + \" failed during API call: \" + response.content);\n }\n\n try {\n const parsedOutput = JSON.parse(response.content);\n if (agent.name === \"SchemaValidator\" && parsedOutput.error) {\n serverLog(\"WARN\", \"SchemaValidator reported validation failure\", { error: parsedOutput.error });\n workflowState[agent.outputKey] = parsedOutput;\n // Using concat\n throw new Error(\"SchemaValidator failed: \" + parsedOutput.error);\n }\n\n workflowState[agent.outputKey] = parsedOutput;\n // Using concat\n serverLog(\"INFO\", \"Workflow Step: \" + agent.name + \" completed successfully.\");\n if (updateCallback) await updateCallback({ status: \"Completed\", agent: agent.name });\n } catch (parseOrValidationError) {\n const errMsg = parseOrValidationError instanceof Error\n ? parseOrValidationError.message\n : String(parseOrValidationError);\n // Using concat\n serverLog(\"ERROR\", \"Error processing output from \" + agent.name, {\n contentPreview: response.content.substring(0, 200) + \"...\",\n error: errMsg,\n });\n if (updateCallback)\n await updateCallback({ status: \"Failed\", agent: agent.name, error: true, message: errMsg });\n // Using concat\n throw new Error(\"Agent \" + agent.name + \" output error: \" + errMsg);\n }\n } // End agent loop\n\n serverLog(\"INFO\", \"Data Generation Workflow Completed Successfully.\");\n if (!workflowState.finalData) throw new Error(\"Workflow completed but final data is missing.\");\n return workflowState.finalData;\n } catch (error) {\n const errorMessage = error instanceof Error ? error.message : String(error);\n serverLog(\"ERROR\", \"Data Generation Workflow Failed\", {\n failedAgent: agents.find(a =>\n workflowState[a.outputKey] === null || (a.outputKey === \"validatedData\" && workflowState.validatedData?.error)\n )?.name || \"Workflow Error\",\n error: errorMessage,\n });\n if (updateCallback) await updateCallback({ status: \"Failed\", error: true, message: errorMessage });\n throw error;\n }\n }\n\n // --- Request Routing ---\n const url = new URL(request.url);\n const pathSegments = url.pathname.split(\"/\").filter(Boolean);\n const method = request.method;\n\n // --- Route: / (Serve HTML Page with Client-Side JS) ---\n if (pathSegments.length === 0 && method === \"GET\") {\n serverLog(\"INFO\", \"Serving root HTML page.\");\n // Client-side JS within the HTML uses template literals, which is fine for browsers\n const htmlResponse = `\n\n\n \n Dynamic Data API\n \n \n \n\n\n

Agent System Interface

\n\n

Multi-Agent Data API

Enter a topic path (e.g., cars/makes/toyota) below or click an example. The system streams agent progress updates.

\n \n

Results

\n Initializing...\n

\n {/* JSON will be rendered here */}\n

Example Topics:

\n {/* Categories and Links */}\n

Science

\n Quantum Entanglement\n Event Horizons\n CRISPR\n

Technology

\n LLMs\n Autonomous Vehicles\n Blockchain Finance\n

\n\n

React Chat Agent

\n {/* React App will render here */}\n

Loading React Chat App...

\n\n

\n\n \n\n`;\n return new Response(htmlResponse, {\n headers: { \"Content-Type\": \"text/html; charset=utf-8\" }, // Ensure UTF-8\n });\n }\n\n // --- Route: /process (AgentA - Chat Interaction, Streaming JSON) ---\n if (pathSegments[0] === \"process\" && method === \"POST\") {\n try {\n const requestData = await request.json();\n const { messages } = requestData;\n const shouldStream = requestData?.stream === true;\n\n if (!messages || !Array.isArray(messages) || messages.length === 0) {\n throw new Error(\"Invalid request: 'messages' array is missing or empty.\");\n }\n serverLog(\"INFO\", \"Processing agent request /process\", {\n messageCount: messages.length,\n streaming: shouldStream,\n });\n\n // Using concat for prompt\n const agentAPrompt =\n \"You are AgentA, an AI assistant. Based on the LATEST user message and conversation history, decide the action:\\n\"\n + \"1. **direct_response**: If you can answer directly (chatting, general knowledge, opinions).\\n\"\n + \"2. **use_tool**: If the user asks for specific, structured data that needs looking up (products, details, specific facts).\\n\\n\"\n + \"Conversation History:\\n\"\n + messages.map(m => m.role + \": \" + m.content).join(\"\\n\") + \"\\n\\n\"\n + \"Current Date/Time: \" + new Date().toISOString() + \"\\n\\n\"\n + \"Output ONLY a single JSON object matching ONE structure:\\n\\n\"\n + \"For direct response:\\n\"\n + \"{\\n\"\n + \" \\\"actionType\\\": \\\"direct_response\\\",\\n\"\n + \" \\\"reasoning\\\": \\\"Why a direct response is suitable.\\\",\\n\"\n + \" \\\"response\\\": \\\"Your textual answer to the user.\\\"\\n\"\n + \"}\\n\\n\"\n + \"For using a tool:\\n\"\n + \"{\\n\"\n + \" \\\"actionType\\\": \\\"use_tool\\\",\\n\"\n + \" \\\"reasoning\\\": \\\"Why a tool is needed and what info you need.\\\",\\n\"\n + \" \\\"response\\\": \\\"Inform the user you're fetching data (e.g., 'Let me look that up...').\\\",\\n\"\n + \" \\\"tool\\\": {\\n\"\n + \" \\\"id\\\": null, \\n\"\n + \" \\\"schema\\\": { \\\"name\\\": \\\"information_lookup\\\", \\\"description\\\": \\\"Schema for the structured data needed.\\\", \\\"fields\\\": [{\\\"name\\\": \\\"field1\\\", \\\"type\\\": \\\"string\\\", \\\"description\\\": \\\"...\\\"}, /* ... more fields */] },\\n\"\n + \" \\\"query\\\": \\\"Specific query for the tool based on user request and schema.\\\",\\n\"\n + \" \\\"parameters\\\": {}\\n\"\n + \" }\\n\"\n + \"}\\n\\n\"\n + \"Ensure the JSON is valid. Focus on the LATEST user message.\";\n\n const lastUserMessageContent = messages[messages.length - 1]?.content || \"[No message content]\";\n\n if (shouldStream) {\n const stream = await callOpenAI(agentAPrompt, lastUserMessageContent, \"gpt-4o-mini\", true, true);\n if (!(stream instanceof ReadableStream)) throw new Error(\"Expected ReadableStream for streaming.\");\n\n // Simple pass-through transform\n const transformStream = new TransformStream({\n transform(chunk, controller) {\n controller.enqueue(chunk);\n },\n flush() {\n serverLog(\"INFO\", \"/process stream flush completed.\");\n },\n });\n stream.pipeTo(transformStream.writable).catch(err =>\n serverLog(\"ERROR\", \"/process stream piping error\", { error: (err as Error).message })\n );\n\n return new Response(transformStream.readable, {\n headers: {\n \"Content-Type\": \"application/json; charset=utf-8\",\n \"Transfer-Encoding\": \"chunked\",\n \"Access-Control-Allow-Origin\": \"*\",\n },\n });\n } else {\n const agentAResponse = await callOpenAI(agentAPrompt, lastUserMessageContent, \"gpt-4o-mini\", true, false);\n if (typeof agentAResponse === \"string\" || agentAResponse instanceof ReadableStream)\n throw new Error(\"Expected structured response for non-streaming.\");\n if (agentAResponse.role === \"system\") {\n // Using concat\n throw new Error(\"AgentA (Process) failed: \" + agentAResponse.content);\n }\n const agentAResult: AgentResponse = JSON.parse(agentAResponse.content);\n if (agentAResult.actionType === \"use_tool\" && agentAResult.tool && !agentAResult.tool.id) {\n agentAResult.tool.id = crypto.randomUUID();\n serverLog(\"INFO\", \"Added UUID to tool request\", { toolId: agentAResult.tool.id });\n }\n serverLog(\"INFO\", \"AgentA response (non-streaming)\", { actionType: agentAResult.actionType });\n return new Response(JSON.stringify(agentAResult), {\n headers: { \"Content-Type\": \"application/json; charset=utf-8\", \"Access-Control-Allow-Origin\": \"*\" },\n });\n }\n } catch (error) {\n const errorMessage = error instanceof Error ? error.message : String(error);\n serverLog(\"ERROR\", \"Agent processing /process failed\", { error: errorMessage });\n const errorResponse: Partial & { error: string; details?: string } = {\n actionType: \"direct_response\",\n // Using concat\n response: \"Processing Error: \" + errorMessage,\n error: \"Agent processing failed\",\n details: errorMessage,\n };\n return new Response(JSON.stringify(errorResponse), {\n status: 500,\n headers: { \"Content-Type\": \"application/json; charset=utf-8\", \"Access-Control-Allow-Origin\": \"*\" },\n });\n }\n }\n\n // --- Route: /tool (AgentB - Execute Tool, Non-Streaming JSON) ---\n if (pathSegments[0] === \"tool\" && method === \"POST\") {\n try {\n const toolRequest: ToolRequest = await request.json();\n if (!toolRequest || !toolRequest.schema || !toolRequest.query)\n throw new Error(\"Invalid tool request. Missing schema or query.\");\n serverLog(\"INFO\", \"Processing tool request /tool\", { toolName: toolRequest.schema.name, toolId: toolRequest.id });\n\n // Using concat for prompt\n const agentBPrompt = \"You are AgentB, executing a tool request.\\n\"\n + \"Tool Schema:\\n\" + JSON.stringify(toolRequest.schema, null, 2) + \"\\n\\n\"\n + \"Query: \" + toolRequest.query + \"\\n\"\n + \"Parameters: \" + JSON.stringify(toolRequest.parameters || {}) + \"\\n\"\n + \"Current Date/Time: \" + new Date().toISOString() + \"\\n\\n\"\n + \"Generate realistic data matching schema based *only* on query/parameters.\\n\"\n + \"Output JSON strictly: { \\\"success\\\": true/false, \\\"data\\\": { /* data */ }, \\\"error\\\": \\\"error msg if false\\\" }\\n\"\n + \"Output ONLY raw JSON.\";\n\n const agentBResponse = await callOpenAI(agentBPrompt, toolRequest.query, \"gpt-4o-mini\", true, false);\n if (typeof agentBResponse === \"string\" || agentBResponse instanceof ReadableStream)\n throw new Error(\"Expected structured response for tool.\");\n if (agentBResponse.role === \"system\") {\n // Using concat\n throw new Error(\"AgentB (Tool) failed: \" + agentBResponse.content);\n }\n\n const toolResponse: ToolResponse = JSON.parse(agentBResponse.content);\n if (typeof toolResponse.success !== \"boolean\") {\n serverLog(\"WARN\", \"Tool AI response missing/invalid 'success'.\", { response: agentBResponse.content });\n toolResponse.success = false;\n toolResponse.error = toolResponse.error || \"AI response invalid structure\";\n }\n if (!toolResponse.success && !toolResponse.error) toolResponse.error = \"Tool failed: unspecified reason from AI.\";\n serverLog(\"INFO\", \"Tool request processed\", { success: toolResponse.success, toolId: toolRequest.id });\n return new Response(JSON.stringify(toolResponse), {\n status: 200,\n headers: { \"Content-Type\": \"application/json; charset=utf-8\", \"Access-Control-Allow-Origin\": \"*\" },\n });\n } catch (error) {\n const errorMessage = error instanceof Error ? error.message : String(error);\n serverLog(\"ERROR\", \"Tool processing /tool failed\", { error: errorMessage });\n return new Response(JSON.stringify({ success: false, error: \"Tool processing error\", details: errorMessage }), {\n status: 500,\n headers: { \"Content-Type\": \"application/json; charset=utf-8\", \"Access-Control-Allow-Origin\": \"*\" },\n });\n }\n }\n\n // --- Route: /path/to/topic (Multi-Agent Data Generation, Streaming Updates & JSON) ---\n if (\n pathSegments.length > 0 && pathSegments[0] !== \"process\" && pathSegments[0] !== \"tool\"\n && pathSegments[0] !== \"favicon.ico\"\n ) {\n const topic = pathSegments.join(\" \");\n const pathString = pathSegments.join(\"/\");\n serverLog(\"INFO\", \"Dynamic data endpoint request (Multi-Agent Streaming)\", { path: pathString, topic });\n\n const transformStream = new TransformStream();\n const writer = transformStream.writable.getWriter();\n const encoder = new TextEncoder();\n\n const sendUpdate = async (update: { status: string; agent?: string; error?: boolean; message?: string }) => {\n try {\n const updateString = JSON.stringify(update) + \"\\n\";\n await writer.write(encoder.encode(updateString));\n } catch (e) {\n serverLog(\"WARN\", \"Failed write update to stream\", { error: (e as Error).message });\n }\n };\n\n // Run workflow async\n (async () => {\n try {\n const finalData = await runDataGenerationWorkflow(topic, pathString, sendUpdate);\n await writer.write(encoder.encode(JSON.stringify(finalData) + \"\\n\"));\n serverLog(\"INFO\", \"Dynamic endpoint success (Multi-Agent Streaming)\", { path: pathString });\n } catch (error) {\n const errorMessage = error instanceof Error ? error.message : String(error);\n serverLog(\"ERROR\", \"Dynamic endpoint failure (Multi-Agent Streaming)\", {\n path: pathString,\n error: errorMessage,\n });\n await sendUpdate({ status: \"Failed\", error: true, message: \"Workflow failed: \" + errorMessage }); // Using concat\n const errorResponse = { error: \"Failed to generate data.\", details: errorMessage, path: \"/\" + pathString }; // Using concat\n await writer.write(encoder.encode(JSON.stringify(errorResponse) + \"\\n\"));\n } finally {\n try {\n await writer.close();\n }\n catch (e) {\n serverLog(\"WARN\", \"Error closing stream writer\", { error: (e as Error).message });\n }\n }\n })();\n\n return new Response(transformStream.readable, {\n status: 200,\n headers: {\n \"Content-Type\": \"application/json; charset=utf-8\",\n \"Transfer-Encoding\": \"chunked\",\n \"Access-Control-Allow-Origin\": \"*\",\n \"Cache-Control\": \"no-cache\",\n },\n });\n }\n\n // --- Fallback: 404 Not Found ---\n serverLog(\"WARN\", \"Route not found\", { path: url.pathname, method });\n return new Response(\n JSON.stringify({\n error: \"Not Found\",\n // Using concat for the final message string\n message: \"The requested path '\" + url.pathname + \"' was not found or the method \" + method\n + \" is not supported for this path.\",\n }),\n {\n status: 404,\n headers: { \"Content-Type\": \"application/json; charset=utf-8\", \"Access-Control-Allow-Origin\": \"*\" },\n },\n );\n} // End of server function","deleted","label",[98],"permalink","https://www.val.town/x/Get/json/code/main.tsx","extension","actionData","errors"]