\n\n

TownCar

Your personal driver for any task.
Tell us the destination, we'll handle the journey.

\n \n

\n\n

Your Itinerary

\n\n \n \n \n\n \n \n \n `;\n}\n\n// --- BACKEND: The Hono API ---\nconst app = new Hono();\nconst openai = new OpenAI();\n\nasync function getOpenAIResponse(systemPrompt, userMessages, jsonResponse = false) {\n try {\n const response = await openai.chat.completions.create({\n model: \"gpt-4-turbo\",\n messages: [\n { role: \"system\", content: systemPrompt },\n ...userMessages,\n ],\n response_format: jsonResponse ? { \"type\": \"json_object\" } : undefined,\n });\n return response.choices[0].message.content;\n } catch (e) {\n console.error(\"OpenAI API error:\", e.message);\n throw new Error(\"Failed to get response from AI model.\");\n }\n}\n\nasync function getOpenAIStream(systemPrompt, luggage) {\n let finalPrompt = systemPrompt;\n for (const key in luggage) {\n finalPrompt = finalPrompt.replace(`{{${key}}}`, luggage[key]);\n }\n\n return openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: [{ role: \"system\", content: finalPrompt }],\n stream: true,\n });\n}\n\n// Helper to fill prompts\nconst fillPrompt = (template, data) => {\n return template.replace(/{{(\\w+)}}/g, (match, key) => data[key] || match);\n};\n\n// Route: /plan\napp.post(\"/plan\", async (c) => {\n const { goal } = await c.req.json();\n if (!goal) return c.json({ error: \"Goal is required\" }, 400);\n\n const plannerPrompt = fillPrompt(META_PROMPTS.ROUTE_PLANNER, { goal });\n try {\n let itineraryJson = await getOpenAIResponse(plannerPrompt, [], true);\n let itinerary = JSON.parse(itineraryJson);\n\n // Validate and retry if necessary\n if (!itinerary.stops || itinerary.stops.length === 0 || itinerary.stops.length > 5) {\n const retryPrompt = fillPrompt(META_PROMPTS.ROUTE_PLANNER_RETRY, { goal });\n itineraryJson = await getOpenAIResponse(retryPrompt, [], true);\n itinerary = JSON.parse(itineraryJson);\n }\n\n return c.json(itinerary);\n } catch (e) {\n console.error(\"Planning Error:\", e);\n return c.json({ details: \"The AI strategist is on a coffee break. Please try again.\" }, 500);\n }\n});\n\n// Route: /drive\napp.post(\"/drive\", async (c) => {\n const { task, context, goal } = await c.req.json();\n\n try {\n // 1. Get the task specification from STOP_ARCHITECT\n const specPrompt = fillPrompt(META_PROMPTS.STOP_ARCHITECT, { context, task });\n const specJson = await getOpenAIResponse(specPrompt, [], true);\n const spec = JSON.parse(specJson);\n\n // 2. Get the UI Schema from UI_BUILDER (if needed)\n let uiSchema = { inputs: [] };\n if (spec.inputs && spec.inputs.length > 0) {\n const uiPrompt = fillPrompt(META_PROMPTS.UI_BUILDER, { spec: specJson });\n const uiSchemaJson = await getOpenAIResponse(uiPrompt, [], true);\n uiSchema = JSON.parse(uiSchemaJson);\n }\n\n // 3. Get the final execution prompt from LOGIC_WEAVER\n const logicPrompt = fillPrompt(META_PROMPTS.LOGIC_WEAVER, { context, spec: specJson });\n const prompt = await getOpenAIResponse(logicPrompt, []);\n\n return c.json({ spec, uiSchema, prompt });\n } catch (e) {\n console.error(\"Drive step error:\", e);\n return c.json({ details: `Failed to process step: ${task}.` }, 500);\n }\n});\n\n// Route: /execute (for streaming)\napp.post(\"/execute\", async (c) => {\n const { prompt, luggage } = await c.req.json();\n try {\n const stream = await getOpenAIStream(prompt, luggage);\n return new Response(stream.toReadableStream(), {\n headers: { \"Content-Type\": \"text/event-stream\" },\n });\n } catch (e) {\n console.error(\"Execution error:\", e);\n return c.json({ details: \"The AI driver ran into an unexpected issue.\" }, 500);\n }\n});\n\n// Route: /\napp.get(\"/\", (c) => {\n return c.html(generateHtml(c.req.url));\n});\n\nexport default app.fetch;","deleted","label","towncar2",[98],"permalink","https://www.val.town/x/join/TownCar/code/main.tsx","extension","actionData","errors"]