FlagCheck analyses chat screenshots you choose to upload and returns an assessment of the conversation. This policy explains exactly what leaves your phone, who else processes it, and what we keep.
Only the images you explicitly select and submit. Before they are sent, the app applies the redactions you chose — cropping the header and blacking out any areas you tapped — and those edits are baked into the image data itself on your device, not merely drawn over the top. The original, unedited screenshot never leaves your phone.
A random identifier such as FC-A3K9QZ-ANON, generated on your device the first
time you open the app. It is not derived from your phone number, your hardware, your accounts
or anything else about you, and it cannot be traced back to you. We use it to apply usage
limits and to track your free-scan balance.
If the app crashes, we receive the error message and stack trace. If you tap the thumbs up/down on an audit, we receive that verdict along with the audit's reference number. Both are filtered before sending and again on arrival to strip anything resembling image data or chat content. They are written to a server log for a human to read and are not stored in any database.
This is the part that matters most. To analyse a conversation, your screenshot is sent from our server to OpenRouter, which routes it to the AI model that performs the analysis. OpenRouter is a third-party company based in the United States, so your screenshot is processed outside your country.
We send only the redacted image, the relationship context you selected, and your chosen language. We do not send your device reference, and there is nothing in the request that identifies you.
| Who | What they receive | Why |
|---|---|---|
| Our server (Vercel) | The redacted screenshot, in memory only | To forward it for analysis |
| OpenRouter | The redacted screenshot | To run the AI analysis |
| RevenueCat | An anonymous purchase identifier | To remember that you paid |
| Google Play | Your payment, handled entirely by Google | We never see card details |
Since we cannot identify you, we also cannot look up, export or delete "your" data on request — there is no record tied to you to act on. The controls above are the complete set, and they are all in your hands.
FlagCheck is not directed at children under 13, and we do not knowingly process their data. The app discusses relationship dynamics and is intended for an adult and teenage audience in line with its store rating.
All traffic between the app, our server and the AI provider is encrypted in transit. Audit results on your device are held in encrypted storage. Our provider credentials live only in server environment variables and are never included in the app.
No system is perfectly secure. The strongest protection available to you is the redaction step: anything you black out before sending never reaches us at all.
If this policy changes materially we will update the date above and note the change in the app. Continuing to use FlagCheck after a change means you accept the revised policy.
Questions about this policy: hmtechinovation@gmail.com
FlagCheck menganalisis screenshot chat yang Anda pilih untuk diunggah, lalu memberikan penilaian atas percakapan tersebut. Kebijakan ini menjelaskan persis apa yang keluar dari ponsel Anda, siapa lagi yang memprosesnya, dan apa yang kami simpan.
Hanya gambar yang Anda pilih dan kirimkan sendiri. Sebelum dikirim, aplikasi menerapkan sensor yang Anda tentukan — memotong bagian header dan menghitamkan area yang Anda ketuk — dan perubahan itu dibakar ke dalam data gambar di perangkat Anda, bukan sekadar ditimpa di layar. Screenshot asli yang belum disensor tidak pernah meninggalkan ponsel Anda.
Kode acak seperti FC-A3K9QZ-ANON, dibuat di perangkat Anda saat pertama kali
membuka aplikasi. Kode ini tidak berasal dari nomor telepon, perangkat keras, akun, atau apa
pun tentang diri Anda, dan tidak bisa dilacak balik ke Anda. Kami memakainya untuk membatasi
penggunaan dan mencatat sisa kuota scan gratis.
Jika aplikasi mengalami crash, kami menerima pesan error dan jejak teknisnya. Jika Anda menekan tombol jempol pada sebuah audit, kami menerima penilaian itu beserta nomor referensi auditnya. Keduanya disaring sebelum dikirim dan disaring lagi saat tiba, untuk membuang apa pun yang menyerupai data gambar atau isi chat. Semuanya ditulis ke log server untuk dibaca manusia, dan tidak disimpan di basis data mana pun.
Ini bagian yang paling penting. Untuk menganalisis percakapan, screenshot Anda dikirim dari server kami ke OpenRouter, yang meneruskannya ke model AI pelaksana analisis. OpenRouter adalah perusahaan pihak ketiga yang berbasis di Amerika Serikat, jadi screenshot Anda diproses di luar negara Anda.
Yang kami kirim hanya gambar yang sudah disensor, konteks hubungan yang Anda pilih, dan bahasa yang Anda pilih. Kode perangkat Anda tidak ikut dikirim, dan tidak ada apa pun dalam permintaan itu yang mengidentifikasi Anda.
| Siapa | Yang mereka terima | Untuk apa |
|---|---|---|
| Server kami (Vercel) | Screenshot tersensor, hanya di memori | Meneruskan untuk dianalisis |
| OpenRouter | Screenshot tersensor | Menjalankan analisis AI |
| RevenueCat | Kode pembelian anonim | Mengingat bahwa Anda sudah membayar |
| Google Play | Pembayaran Anda, sepenuhnya ditangani Google | Kami tidak pernah melihat data kartu |
Karena kami tidak bisa mengidentifikasi Anda, kami juga tidak bisa mencari, mengekspor, atau menghapus data "milik Anda" atas permintaan — tidak ada catatan yang terikat pada Anda untuk ditindak. Kontrol di atas adalah keseluruhannya, dan semuanya ada di tangan Anda.
FlagCheck tidak ditujukan untuk anak di bawah 13 tahun, dan kami tidak dengan sengaja memproses data mereka. Aplikasi ini membahas dinamika hubungan dan ditujukan untuk kalangan dewasa dan remaja sesuai rating toko aplikasinya.
Seluruh lalu lintas antara aplikasi, server kami, dan penyedia AI dienkripsi saat pengiriman. Hasil audit di perangkat Anda disimpan terenkripsi. Kredensial penyedia hanya ada di variabel lingkungan server dan tidak pernah disertakan dalam aplikasi.
Tidak ada sistem yang sepenuhnya aman. Perlindungan terkuat yang tersedia bagi Anda adalah langkah sensor: apa pun yang Anda hitamkan sebelum mengirim tidak pernah sampai kepada kami.
Jika kebijakan ini berubah secara material, kami akan memperbarui tanggal di atas dan memberi tahu di aplikasi. Melanjutkan penggunaan FlagCheck setelah perubahan berarti Anda menerima kebijakan yang telah direvisi.
Pertanyaan tentang kebijakan ini: hmtechinovation@gmail.com
FlagCheck analiza las capturas de pantalla de conversaciones que decides subir y devuelve una valoración de la conversación. Esta política explica exactamente qué sale de tu teléfono, quién más lo procesa y qué guardamos.
Solo las imágenes que seleccionas y envías de forma explícita. Antes de enviarlas, la aplicación aplica las censuras que hayas elegido — recortar la cabecera y tapar en negro las zonas que tocaste — y esos cambios quedan grabados en los datos de la imagen en tu dispositivo, no simplemente dibujados por encima. La captura original sin censurar nunca sale de tu teléfono.
Un identificador aleatorio como FC-A3K9QZ-ANON, generado en tu dispositivo la
primera vez que abres la aplicación. No deriva de tu número de teléfono, tu hardware, tus
cuentas ni de nada relativo a ti, y no puede rastrearse hasta ti. Lo usamos para aplicar
límites de uso y llevar la cuenta de tus análisis gratuitos.
Si la aplicación falla, recibimos el mensaje de error y su traza técnica. Si pulsas el pulgar arriba o abajo en una auditoría, recibimos ese veredicto junto con el número de referencia de la auditoría. Ambos se filtran antes de enviarse y de nuevo al llegar, para eliminar cualquier cosa que se parezca a datos de imagen o al contenido de la conversación. Se escriben en un registro del servidor para que una persona los lea y no se guardan en ninguna base de datos.
Esta es la parte más importante. Para analizar una conversación, tu captura se envía desde nuestro servidor a OpenRouter, que la dirige al modelo de IA que realiza el análisis. OpenRouter es una empresa externa con sede en Estados Unidos, así que tu captura se procesa fuera de tu país.
Solo enviamos la imagen ya censurada, el contexto de relación que seleccionaste y el idioma que elegiste. No enviamos la referencia de tu dispositivo, y no hay nada en la petición que te identifique.
| Quién | Qué recibe | Para qué |
|---|---|---|
| Nuestro servidor (Vercel) | La captura censurada, solo en memoria | Reenviarla para su análisis |
| OpenRouter | La captura censurada | Ejecutar el análisis de IA |
| RevenueCat | Un identificador de compra anónimo | Recordar que has pagado |
| Google Play | Tu pago, gestionado íntegramente por Google | Nunca vemos los datos de tu tarjeta |
Dado que no podemos identificarte, tampoco podemos localizar, exportar ni borrar «tus» datos si lo solicitas: no existe ningún registro vinculado a ti sobre el que actuar. Los controles anteriores son el conjunto completo, y están todos en tus manos.
FlagCheck no está dirigida a menores de 13 años y no procesamos sus datos de forma consciente. La aplicación trata dinámicas de relación y está destinada a un público adulto y adolescente, en línea con su clasificación en la tienda.
Todo el tráfico entre la aplicación, nuestro servidor y el proveedor de IA va cifrado en tránsito. Los resultados en tu dispositivo se guardan cifrados. Las credenciales del proveedor residen únicamente en variables de entorno del servidor y nunca se incluyen en la aplicación.
Ningún sistema es perfectamente seguro. La protección más fuerte a tu alcance es el paso de censura: lo que tapes antes de enviar no llega nunca hasta nosotros.
Si esta política cambia de forma sustancial, actualizaremos la fecha de arriba y lo indicaremos en la aplicación. Seguir usando FlagCheck después de un cambio implica que aceptas la política revisada.
Dudas sobre esta política: hmtechinovation@gmail.com