🛡 SafeSense Guardian – AI Content Moderator

\n Analyze text, images and documents as safe or unsafe{\" \"}\n with category, sensitivity score and explanation.\n

\n\n

\n setMode(\"text\")}\n style={{\n margin: \"5px\",\n padding: \"8px 14px\",\n backgroundColor: mode === \"text\" ? \"#007bff\" : \"#cccccc\",\n color: \"white\",\n border: \"none\",\n borderRadius: \"5px\",\n cursor: \"pointer\",\n }}\n >\n ✏ Text\n \n setMode(\"image\")}\n style={{\n margin: \"5px\",\n padding: \"8px 14px\",\n backgroundColor: mode === \"image\" ? \"#007bff\" : \"#cccccc\",\n color: \"white\",\n border: \"none\",\n borderRadius: \"5px\",\n cursor: \"pointer\",\n }}\n >\n 🖼 Image\n \n setMode(\"document\")}\n style={{\n margin: \"5px\",\n padding: \"8px 14px\",\n backgroundColor: mode === \"document\" ? \"#007bff\" : \"#cccccc\",\n color: \"white\",\n border: \"none\",\n borderRadius: \"5px\",\n cursor: \"pointer\",\n }}\n >\n 📄 Document\n \n

\n\n {/* TEXT MODE */}\n {mode === \"text\" && (\n

\n setTextInput(e.target.value)}\n placeholder=\"Paste or type text to moderate...\"\n style={{ width: \"100%\", height: \"150px\", padding: \"10px\" }}\n />\n moderateContent(\"text\")}\n disabled={!textInput || loading}\n >\n {loading ? \"Analyzing...\" : \"Analyze Text\"}\n \n

\n )}\n\n {/* IMAGE MODE */}\n {mode === \"image\" && (\n

\n \n \n \n \n 🖼 Capture Frame\n \n moderateContent(\"image\")}\n disabled={!selectedImage || loading}\n >\n {loading ? \"Analyzing...\" : \"Analyze Image\"}\n \n\n {selectedImage && (\n

\n \n

\n )}\n\n \n \n

\n )}\n\n {/* DOCUMENT MODE */}\n {mode === \"document\" && (\n

\n \n \n {selectedFileName && (\n

\n Selected: {selectedFileName}\n

\n )}\n

\n Demo: file is treated as a small text blob and moderated as text.\n

\n )}\n\n {/* RESULT */}\n {result && (\n \n

Moderation Result

\n Label: {result.label}\n

\n Category: {result.category}\n

\n Sensitivity: {Number(result.sensitivity).toFixed(2)}\n

\n Reason: {result.reason}\n

\n \n )}\n\n

\n This demo respects intellectual property and only analyzes content for\n moderation purposes.\n

\n\n \n Open this val\n \n \n );\n}\n\nfunction client() {\n createRoot(document.getElementById(\"root\")!).render();\n}\nif (typeof document !== \"undefined\") {\n client();\n}\n\n// ============ BACKEND (HTTP val) ============\n\nexport default async function server(request: Request): Promise {\n if (\n request.method === \"POST\" && new URL(request.url).pathname === \"/moderate\"\n ) {\n try {\n const { OpenAI } = await import(\"https://esm.town/v/std/openai\"); // Val Town built‑in proxy [web:26][web:36]\n const openai = new OpenAI();\n\n const body = await request.json();\n const type = body.type as \"text\" | \"image\" | \"document\";\n\n let description = \"\";\n let label = \"safe\";\n let category = \"neutral\";\n let sensitivity = 0.0;\n let reason = \"\";\n\n // Common helper prompt\n const systemPrompt =\n \"You are SafeSense Guardian, an AI content moderator. \" +\n \"Classify the content as safe or unsafe, with category in {nsfw, violence, neutral, other}. \" +\n \"Return a short JSON object ONLY with keys label, category, sensitivity (0-1), reason. \" +\n \"Reason must briefly explain why it is safe/unsafe.\";\n\n if (type === \"text\") {\n const text = String(body.text || \"\");\n const resp = await openai.chat.completions.create({\n model: \"gpt-4o-mini\",\n messages: [\n { role: \"system\", content: systemPrompt },\n {\n role: \"user\",\n content: `Moderate this text:\\n\\n${text}`,\n },\n ],\n max_tokens: 200,\n });\n\n const raw = resp.choices[0].message.content || \"{}\";\n try {\n const parsed = JSON.parse(raw as string);\n label = parsed.label ?? label;\n category = parsed.category ?? category;\n sensitivity = parsed.sensitivity ?? sensitivity;\n reason = parsed.reason ?? \"\";\n } catch {\n reason = \"Model returned non‑JSON content.\";\n }\n } else if (type === \"image\") {\n const imageDataUrl: string = body.image;\n // Expect a full data URL: data:image/jpeg;base64,xxx [web:32][web:38]\n const resp = await openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: [\n { role: \"system\", content: systemPrompt },\n {\n role: \"user\",\n content: [\n {\n type: \"text\",\n text: \"Moderate this image and return JSON only.\",\n },\n {\n type: \"image_url\",\n image_url: {\n url: imageDataUrl,\n },\n },\n ],\n },\n ],\n max_tokens: 200,\n });\n\n const raw = resp.choices[0].message.content || \"{}\";\n try {\n const parsed = JSON.parse(raw as string);\n label = parsed.label ?? label;\n category = parsed.category ?? category;\n sensitivity = parsed.sensitivity ?? sensitivity;\n reason = parsed.reason ?? \"\";\n } catch {\n reason = \"Model returned non‑JSON content for image.\";\n }\n } else if (type === \"document\") {\n // For demo, treat document data URL as “pseudo text” and ask model to\n // infer likely content risk (no real PDF parsing to respect sandbox). [web:40]\n const docData = String(body.document || \"\");\n const resp = await openai.chat.completions.create({\n model: \"gpt-4o-mini\",\n messages: [\n { role: \"system\", content: systemPrompt },\n {\n role: \"user\",\n content: \"A user uploaded a small document. \" +\n \"You see only a short snippet of its textual content (or a description of the file). \" +\n \"Moderate based on this limited information:\\n\\n\" +\n docData.slice(0, 4000),\n },\n ],\n max_tokens: 200,\n });\n\n const raw = resp.choices[0].message.content || \"{}\";\n try {\n const parsed = JSON.parse(raw as string);\n label = parsed.label ?? label;\n category = parsed.category ?? category;\n sensitivity = parsed.sensitivity ?? sensitivity;\n reason = parsed.reason ?? \"\";\n } catch {\n reason = \"Model returned non‑JSON content for document.\";\n }\n } else {\n return new Response(\n JSON.stringify({\n label: \"unsafe\",\n category: \"other\",\n sensitivity: 1.0,\n reason: \"Unsupported content type.\",\n }),\n { status: 400, headers: { \"Content-Type\": \"application/json\" } },\n );\n }\n\n description = `Content classified as ${label} in category ${category}.`;\n\n return new Response(\n JSON.stringify({\n label,\n category,\n sensitivity,\n reason: reason || description,\n }),\n { headers: { \"Content-Type\": \"application/json\" } },\n );\n } catch (e) {\n console.error(\"Moderation server error:\", e);\n return new Response(\n JSON.stringify({\n label: \"unsafe\",\n category: \"technical_error\",\n sensitivity: 1.0,\n reason: \"Recognition / moderation failed on server.\",\n }),\n { status: 500, headers: { \"Content-Type\": \"application/json\" } },\n );\n }\n }\n\n // GET: serve React app\n const css = `\nbody {\n font-family: Arial, sans-serif;\n background-color: #e9ecef;\n margin: 0;\n padding: 20px;\n}\nbutton {\n margin: 8px;\n padding: 8px 14px;\n background-color: #007bff;\n color: white;\n border: none;\n border-radius: 5px;\n cursor: pointer;\n}\nbutton:hover {\n background-color: #0056b3;\n}\ntextarea {\n font-family: inherit;\n}\n`;\n\n return new Response(\n `\n \n \n SafeSense Guardian – AI Content Moderator\n \n \n \n \n

\n \n \n \n \n `,\n {\n headers: { \"Content-Type\": \"text/html\" },\n },\n );\n}","deleted","label",[99],"permalink","https://www.val.town/x/mprasadhv/untitled-2950/code/main.ts","extension","actionData","errors"]