Reply Forge

Paste any X post → get a sharp, factual 50-word reply. No restrictions.

\n\n

🔑 Add your Grok API key

Get your API key at console.x.ai. Saved only in your browser. Grok has no content restrictions.

\n \n \n

\n\n

Paste tweet text

\n \n

💡 Copy the text from X — URLs alone won't work

\n\n Reply vibe\n

\n \n \n \n \n \n

\n\n \n\n

Crafting your reply…

\n ✦ Your Reply\n —\n

Key facts used

\n \n \n

\n\n

\n\n\n`;\n\nexport default async function handler(req: Request): Promise {\n const url = new URL(req.url);\n\n // Serve the HTML page\n if (req.method === \"GET\") {\n return new Response(HTML, { headers: { \"Content-Type\": \"text/html\" } });\n }\n\n // API proxy endpoint\n if (req.method === \"POST\" && url.pathname === \"/api\") {\n try {\n const { key, input, tone } = await req.json();\n\n if (!key || !key.startsWith(\"xai-\")) {\n return Response.json({ error: \"Invalid Grok API key\" }, {\n status: 400,\n });\n }\n if (!input) {\n return Response.json({ error: \"No input provided\" }, { status: 400 });\n }\n\n const toneMap: Record = {\n energetic:\n \"enthusiastic, punchy, high-energy — active verbs, exclamation where natural\",\n witty: \"clever, light, a touch of humor — smart but not snarky\",\n authoritative:\n \"confident, expert, data-forward — commanding and credible\",\n supportive:\n \"warm, affirming, constructive — backing the person with facts\",\n contrarian:\n \"bold, respectfully challenging — pushback grounded only in the post\",\n };\n\n // Call Grok API (uses OpenAI-compatible endpoint)\n const grokRes = await fetch(\"https://api.x.ai/v1/chat/completions\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${key}`,\n },\n body: JSON.stringify({\n model: \"grok-beta\",\n messages: [\n {\n role: \"system\",\n content:\n `You are a social media reply expert. Write a reply to the X (Twitter) post the user provides.\n\nRULES:\n- Max 50 words. Count every word carefully.\n- Tone: ${toneMap[tone] || toneMap.energetic}\n- Only use facts and claims stated in the post. Do not invent anything.\n- Respond with ONLY a raw JSON object — no markdown, no code fences, no text outside the JSON.\n- Format: {\"reply\":\"your reply text\",\"keyFacts\":[\"fact 1 from post\",\"fact 2\",\"fact 3\"]}\n- If for any reason you cannot write a reply, return: {\"reply\":\"\",\"keyFacts\":[],\"error\":\"brief reason\"}`,\n },\n {\n role: \"user\",\n content: input,\n },\n ],\n temperature: 0.7,\n max_tokens: 400,\n }),\n });\n\n const grokData = await grokRes.json();\n\n if (!grokRes.ok) {\n return Response.json(\n {\n error: grokData?.error?.message ||\n \"Grok API error \" + grokRes.status,\n },\n { status: 502 },\n );\n }\n\n const raw = grokData?.choices?.[0]?.message?.content?.trim() || \"\";\n\n if (!raw) {\n return Response.json({ error: \"Empty response from Grok\" }, {\n status: 502,\n });\n }\n\n // Parse JSON robustly\n let parsed: any = null;\n for (\n const fn of [\n () => JSON.parse(raw),\n () => JSON.parse(raw.replace(/```json|```/gi, \"\").trim()),\n () => {\n const m = raw.match(/\\{[\\s\\S]*\\}/);\n if (m) return JSON.parse(m[0]);\n throw new Error(\"no json\");\n },\n ]\n ) {\n try {\n parsed = fn();\n if (parsed) break;\n } catch {}\n }\n\n // If couldn't parse JSON, Grok responded in plain text\n if (!parsed) {\n return Response.json({\n error: \"Response was not JSON\",\n rawResponse: raw,\n }, { status: 200 });\n }\n\n return Response.json(parsed);\n } catch (e: any) {\n return Response.json({ error: e.message }, { status: 500 });\n }\n }\n\n return new Response(\"Not found\", { status: 404 });\n}","deleted","label",[100],"permalink","https://www.val.town/x/ChiefP/untitled-4667/code/main.ts","extension","actionData","errors"]