🤔 ThinkingFlow MVP 🤔

\n Apply the Six Thinking Hats to your query. Define a hat sequence (e.g., W,R,B,Y,G,B) or use the default.\n
Workflow is visualized in the log below. Current Date: ${currentDate}\n

\n \n

Enter a query and optionally a sequence to start.

\n Powered by Val Town | View Source\n

\n\n \n \n `;\n}\n\n// --- Main Request Handler (Server Code for ThinkingFlow MVP) ---\nexport default async function(req: Request) {\n // Dynamic Import of OpenAI Library\n const { OpenAI } = await import(\"https://esm.town/v/std/openai\");\n\n // --- OpenAI API Call Helper (Reused) ---\n async function callOpenAI(\n systemPrompt: string,\n userMessage: string, // Can be query or intermediate context\n model = \"gpt-4o-mini\",\n ): Promise<{ role: \"assistant\" | \"system\"; content: string }> {\n try {\n const openai = new OpenAI();\n\n const response = await openai.chat.completions.create({\n model: model,\n messages: [\n { role: \"system\", content: systemPrompt },\n { role: \"user\", content: userMessage },\n ],\n temperature: 0.5, // Consistent temperature for structured thinking\n });\n\n if (!response.choices?.[0]?.message?.content) {\n console.error(\"OpenAI API returned unexpected structure:\", JSON.stringify(response));\n throw new Error(\"Received invalid or empty response from AI model.\");\n }\n return { role: \"assistant\", content: response.choices[0].message.content };\n } catch (error) {\n console.error(`OpenAI API call failed for model ${model}. Error:`, error.message, error.response?.data);\n let errorMessage = `Error with AI model (${model}).`;\n let statusCode = error.status || error.response?.status;\n if (statusCode === 401) errorMessage = \"OpenAI Auth Error (401). Check Val Town 'openai' secret.\";\n else if (statusCode === 429) errorMessage = \"OpenAI Rate Limit/Quota Error (429). Check OpenAI plan.\";\n else if (statusCode === 400) errorMessage = `OpenAI Bad Request (400). Details: ${error.message}`;\n else if (statusCode >= 500) errorMessage = `OpenAI Server Error (${statusCode}). Try again later.`;\n else if (error.code === \"ENOTFOUND\" || error.code === \"ECONNREFUSED\")\n errorMessage = `Network Error (${error.code}). Cannot connect to OpenAI.`;\n else if (error.message.includes(\"secret\")) errorMessage = error.message;\n else errorMessage += ` Details: ${error.message}`;\n return { role: \"system\", content: `System Error: ${errorMessage}` };\n }\n }\n\n // --- Helper: Map Hat Code to Prompt and Name ---\n function getHatDetails(hatCode: string): { name: string; prompt: string } {\n switch (hatCode.toUpperCase()) {\n case \"W\":\n return { name: \"White Hat (Facts)\", prompt: whiteHatPrompt };\n case \"R\":\n return { name: \"Red Hat (Emotions)\", prompt: redHatPrompt };\n case \"B\":\n return { name: \"Black Hat (Caution)\", prompt: blackHatPrompt };\n case \"Y\":\n return { name: \"Yellow Hat (Benefits)\", prompt: yellowHatPrompt };\n case \"G\":\n return { name: \"Green Hat (Creativity)\", prompt: greenHatPrompt };\n case \"L\":\n return { name: \"Blue Hat (Summary)\", prompt: blueHatSummaryPrompt }; // 'L' for Blue\n default:\n return { name: \"Unknown Hat\", prompt: \"\" }; // Should not happen with validation\n }\n }\n\n // --- Helper: Run the ThinkingFlow Workflow ---\n async function runThinkingFlow(\n userQuery: string,\n sequenceString: string,\n ): Promise> {\n const conversationLog: Array<{ agent: string; message: string }> = [];\n const defaultSequence = [\"W\", \"R\", \"B\", \"Y\", \"G\", \"L\"]; // L for Blue summary\n const validHatCodes = [\"W\", \"R\", \"B\", \"Y\", \"G\", \"L\"];\n\n // --- Input Validation ---\n if (!userQuery || userQuery.trim().length === 0) {\n conversationLog.push({ agent: \"⚙️ System\", message: \"Input query is empty.\" });\n return conversationLog;\n }\n const trimmedQuery = userQuery.trim();\n if (trimmedQuery.length > 2000) {\n conversationLog.push({ agent: \"👤 User\", message: trimmedQuery.substring(0, 2000) + \"...\" });\n conversationLog.push({ agent: \"⚙️ System\", message: \"Query too long (max 2000 chars).\" });\n return conversationLog;\n }\n conversationLog.push({ agent: \"👤 User\", message: trimmedQuery });\n\n // --- Sequence Processing ---\n let hatSequence: string[] = [];\n let sequenceSource = \"\";\n if (sequenceString && sequenceString.trim().length > 0) {\n hatSequence = sequenceString.toUpperCase().split(\",\").map(s => s.trim()).filter(s => validHatCodes.includes(s));\n if (hatSequence.length > 0) {\n sequenceSource = `Custom sequence: ${hatSequence.join(\", \")}`;\n } else {\n conversationLog.push({\n agent: \"⚙️ System\",\n message: `Invalid custom sequence provided ('${sequenceString}'). Falling back to default.`,\n });\n hatSequence = defaultSequence;\n sequenceSource = `Default sequence: ${hatSequence.join(\", \")}`;\n }\n } else {\n hatSequence = defaultSequence;\n sequenceSource = `Default sequence: ${hatSequence.join(\", \")}`;\n }\n\n // Ensure the *last* step is Blue Hat summary if not explicitly included at end.\n // Or, if L is included, treat it as the final summary step.\n const hasExplicitBlueEnd = hatSequence[hatSequence.length - 1] === \"L\";\n const sequenceToRun = hasExplicitBlueEnd ? hatSequence.slice(0, -1) : hatSequence.filter(h => h !== \"L\"); // Sequence for iteration (exclude final L)\n const finalBlueNeeded = true; // Always run blue summary at the end in this MVP\n\n conversationLog.push({ agent: \"➡️ Workflow\", message: `Starting ThinkingFlow for query: \"${trimmedQuery}\"` });\n conversationLog.push({ agent: \"➡️ Workflow\", message: `Using: ${sequenceSource}` });\n\n // --- Hat Iteration Phase ---\n const collectedOutputs: { [hatName: string]: string } = {};\n let contextForNextHat = \"\"; // Simple context passing (last response)\n\n for (const hatCode of sequenceToRun) {\n const { name, prompt } = getHatDetails(hatCode);\n if (!prompt) continue; // Skip if unknown hat somehow got through\n\n conversationLog.push({ agent: \"➡️ Workflow\", message: `Activating ${name}...` });\n\n // Prepare prompt with query and potentially context\n let currentPrompt = prompt.replace(\"{{USER_QUERY}}\", trimmedQuery);\n // Simple context: provide the last hat's response (optional, can be refined)\n // currentPrompt = currentPrompt.replace(\"{{CONTEXT}}\", contextForNextHat ? `\\nPrevious Hat Output:\\n${contextForNextHat}` : \"\");\n // For MVP, let's keep context simple or omit it for clarity unless specifically needed\n currentPrompt = currentPrompt.replace(\"{{CONTEXT}}\", \"\"); // No inter-hat context for MVP\n\n const hatResponse = await callOpenAI(currentPrompt, trimmedQuery); // Pass original query as user message\n\n // Log response or error\n const agentName = name.split(\" \")[0] + \" Hat\"; // e.g. \"White Hat\"\n conversationLog.push({ agent: agentName, message: hatResponse.content });\n\n if (hatResponse.role === \"assistant\") {\n collectedOutputs[name] = hatResponse.content; // Store successful output\n contextForNextHat = hatResponse.content; // Update context for next potential use\n } else {\n collectedOutputs[name] = `(${name} failed: ${hatResponse.content})`; // Store error\n contextForNextHat = \"\"; // Reset context on error\n }\n }\n conversationLog.push({ agent: \"➡️ Workflow\", message: \"Hat sequence processing complete.\" });\n\n // --- Final Blue Hat Summary Phase ---\n if (finalBlueNeeded) {\n conversationLog.push({ agent: \"➡️ Workflow\", message: `Activating Blue Hat (Summary)...` });\n const { name: blueName, prompt: bluePrompt } = getHatDetails(\"L\");\n\n let outputsText = Object.entries(collectedOutputs)\n .map(([hat, output]) => `--- ${hat} ---\\n${output}`)\n .join(\"\\n\\n\");\n if (!outputsText) outputsText = \"(No outputs were collected from the sequence).\";\n\n const finalPrompt = bluePrompt\n .replace(\"{{USER_QUERY}}\", trimmedQuery)\n .replace(\"{{SEQUENCE}}\", hatSequence.join(\", \")) // Show the full original sequence planned\n .replace(\"{{COLLECTED_OUTPUTS}}\", outputsText);\n\n const summaryResponse = await callOpenAI(finalPrompt, \"Synthesize the collected outputs.\");\n\n conversationLog.push({ agent: \"Blue Hat\", message: summaryResponse.content }); // Log summary or error\n }\n\n conversationLog.push({ agent: \"➡️ Workflow\", message: \"ThinkingFlow Complete.\" });\n\n return conversationLog;\n }\n\n // --- Request Handling Logic ---\n const url = new URL(req.url);\n const format = url.searchParams.get(\"format\");\n const userQuery = url.searchParams.get(\"query\") || \"\";\n const sequenceString = url.searchParams.get(\"sequence\") || \"\"; // Get sequence string\n const sourceUrl = `https://val.town/v/${Deno.env.get(\"val_owner\")}.${Deno.env.get(\"val_slug\")}`;\n\n // API Request (format=json)\n if (format === \"json\") {\n if (!userQuery || userQuery.trim().length === 0) {\n return new Response(JSON.stringify({ error: \"Query parameter is missing\" }), {\n status: 400,\n headers: { \"Content-Type\": \"application/json\", \"Access-Control-Allow-Origin\": \"*\" },\n });\n }\n try {\n const conversationLog = await runThinkingFlow(userQuery, sequenceString);\n return new Response(JSON.stringify(conversationLog), {\n status: 200,\n headers: { \"Content-Type\": \"application/json\", \"Access-Control-Allow-Origin\": \"*\" },\n });\n } catch (error) {\n console.error(\"Critical error during runThinkingFlow:\", error);\n return new Response(\n JSON.stringify({\n error: \"ThinkingFlow processing failed.\",\n details: error.message,\n }),\n {\n status: 500,\n headers: { \"Content-Type\": \"application/json\", \"Access-Control-Allow-Origin\": \"*\" },\n },\n );\n }\n }\n // HTML Page Load\n else {\n // Pass sequence to HTML shell for input field default\n const html = generateHtmlShell(userQuery, sequenceString, sourceUrl);\n return new Response(html, {\n status: 200,\n headers: { \"Content-Type\": \"text/html\" },\n });\n }\n}","deleted","label",[98],"permalink","https://www.val.town/x/Get/Thinking/code/main.tsx","extension","actionData","errors"]