StreamSparkIn development
← Back to StreamSpark

The details, in plain language

Privacy.

Your content is personal. Here’s how the current local version handles it.

Last updated October 11, 2026. StreamSpark is operated by Shehab. For privacy questions or requests, email wonderlast888@gmail.com. This policy covers the website and the StreamSpark app for Windows, which is in private development. By using StreamSpark you agree to this policy.

In short

StreamSpark runs on your own computer. It has no server, user accounts, advertising or analytics of its own, and it never sells or shares your data for advertising. Your footage, clips, drafts, posting records and statistics stay on your computer. Data leaves it only to the services described below, when you use the feature that needs them.

The website

The website has no signup form, cookies of its own, advertising or analytics scripts, and lets no third party serve content on it. Cloudflare hosts it and may process technical request data, including IP addresses, requested URLs, browser information and timestamps, to deliver and protect it, and may set security cookies where its protection features require them. See Cloudflare’s Privacy Policy. Two small pages on the website, for Facebook and Instagram sign-in, are described under Facebook and Instagram.

Watching streams

StreamSpark records the public live streams you add (Kick, Twitch and TikTok LIVE) and reads their public chat to find moments. It uses channel names, profile pictures, stream details, chat messages and reaction counts. A chat log keeps message text and times, with a code in place of each sender (names and IDs are not written), and is deleted after 14 days. Some streams send chat events through a Cloudflare tunnel.

TikTok LIVE chat is read with TikTokLive, an unofficial open-source connector. To connect, it asks the Euler Stream signing service to sign its request to TikTok, so the stream’s address is sent to that service. You can also save a TikTok login (the session cookie values from your own signed-in browser) so StreamSpark can open live streams that TikTok shows only to signed-in viewers. It is optional, stored with Windows account-bound encryption, sent only to tiktok.com to read a stream’s details, never sent to the signing service and never used for chat or posting. Disconnect, in the Add streamer window, deletes it.

Captions by AssemblyAI

When a clip is listened to (for its automatic title, or when you press Generate captions in Studio, including for phone videos), its sound is sent to AssemblyAI over an encrypted connection to turn speech into timed words. The picture, chat and other details are not sent. The words are kept on your computer and reused, so a clip is sent once. As soon as they arrive, StreamSpark asks AssemblyAI to delete the transcript and the uploaded sound; if that fails, it asks again for up to 7 days, and AssemblyAI deletes transcripts on its own after 30 days. AssemblyAI keeps some request metadata for billing and logging, and says it may use submitted audio, with personal information removed, to improve its models unless the account opts out in its Data Controls (paid plans). Captions use only the AssemblyAI account and key you connect in Settings. See AssemblyAI’s Privacy Policy. You can type captions by hand instead.

Titles by Claude

When you connect Claude in Settings, each clip captured afterwards gets a title and a short post description written by Anthropic’s Claude through the Anthropic API. Only text is sent, over an encrypted connection: the clip’s words with rough times, the streamer’s display name and platform, the clip’s length and its hottest moment. The picture, sound and chat are not sent, and neither is any data from your connected accounts. Clips captured before you connect, or while titles are off, are never sent. Anthropic states that it does not train on API inputs and outputs by default and deletes them within 30 days, unless a longer period is agreed, needed to enforce its usage policy or required by law. See Anthropic’s Privacy Policy. If you choose the free model under “Titles written by”, titles are written on your computer and nothing is sent to Anthropic.

Sounds by Epidemic Sound

If you connect Epidemic Sound in Settings, your sound searches and the sounds you play or add are requested from Epidemic Sound with your API key and a random ID for this computer that does not identify you. Sounds are downloaded once and kept on your computer to reuse; music previews are kept for up to an hour. Your own sound files stay on your computer. See Epidemic Sound’s Privacy Policy.

Record with phone

Record with phone works only on your home network. While a phone is paired, StreamSpark opens a small page on your computer’s home-network address. Every request needs the secret code in the QR code, an unscanned code expires after 10 minutes, and Disconnect phone closes it. Videos go straight from your phone to your computer over your Wi-Fi, not through the internet or any server; this local connection is not encrypted, so use it only on a network you trust. Takes wait in the phone page’s own storage until your computer has them. Phone videos stay on your computer until you delete them.

Connected accounts

You connect posting accounts yourself, through each platform’s own sign-in, using developer apps and keys you set up. Brands let you group up to five sets of accounts, one account per platform in each. For every connected account StreamSpark keeps, on your computer: the account’s name, ID and profile picture (refreshed daily and deleted when you disconnect), its sign-in tokens, and a record of each post (the file, its status, the platform’s ID for it and the text you wrote). Tokens and keys are protected with Windows account-bound encryption. Posts are sent only when you confirm them, right away or at a time you schedule. StreamSpark has no server of its own that receives your tokens, videos or captions.

YouTube

StreamSpark uses YouTube API Services. If you connect YouTube, it uses Google sign-in with two permissions. youtube.upload: to upload the finished video you choose, with the details you enter (title, description, audience, visibility, an optional cover picture and an optional publish time). youtube.readonly: to show your channel’s name and picture, check each upload’s processing and visibility, and read the view, like and comment counts of videos uploaded through StreamSpark and your channel’s subscriber count. YouTube data stays on your computer: it is not sent to AssemblyAI, Anthropic or anyone else, is not used for advertising, and is not read by people other than you.

StreamSpark’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements. By using StreamSpark’s YouTube features you agree to be bound by the YouTube Terms of Service. Google handles data under the Google Privacy Policy. You can revoke StreamSpark’s access at any time in Google’s security settings.

TikTok

If you connect TikTok, StreamSpark uses TikTok’s Login Kit, Content Posting API and Display API with four permissions. user.info.basic: your account’s name and profile picture. video.upload: to send the finished video you choose to your TikTok inbox as a draft; you finish and post it inside TikTok. video.list: your public videos’ IDs, titles, post times, lengths and view, like, comment and share counts, used to find the video your draft became and show its numbers in Stats; these details are kept for 100 days after a video was posted. user.info.stats: your follower, like and video counts. StreamSpark also keeps TikTok’s ID for the account, so it can tell whether you reconnect the same one. It does not read your messages, comments, follower list or private videos. While StreamSpark is in TikTok’s testing stage, only approved test accounts can connect. TikTok handles data under its Privacy Policy.

X

If you connect X, StreamSpark uses X’s OAuth 2.0 sign-in with users.read and tweet.read (your account’s name and picture, and the numbers of posts made through StreamSpark), tweet.write and media.write (upload the finished video and post it) and offline.access (stay signed in). It sends the video and the text you write, and the post is public. Stats reads the public numbers of posts made through StreamSpark for their first 28 days (and, for posts under 30 days old, their profile and link clicks) and your follower count. It does not read your timeline, messages or other posts. The X API charges your own X developer account: StreamSpark stops its Stats reads for the day once their estimated cost reaches 25 cents, and stops when X says your credits or spending limit are used up. X handles data under its Privacy Policy.

Facebook and Instagram

If you connect Facebook, StreamSpark uses Facebook Login for Business, with a Meta app that you set up, to list the Pages you manage, post Reels to the Page you choose and read those Reels’ numbers. If an Instagram professional account is linked to that Page, StreamSpark can post Reels to it and read their numbers through the same connection. The login configuration holds pages_show_list, pages_read_engagement, pages_manage_posts, read_insights, business_management (needed when a Page is owned by a business portfolio), instagram_basic, instagram_content_publish and instagram_manage_insights.

An Instagram professional account can instead connect with its own Instagram login, without a Page, using instagram_business_basic, instagram_business_content_publish and instagram_business_manage_insights. Instagram fetches the video itself on this path, so StreamSpark puts a temporary private copy in a Cloudflare R2 storage bucket in your own Cloudflare account, gives Instagram a link that expires after 30 minutes, and deletes the copy once Instagram has it, or after about 2 hours while StreamSpark is running. When cloud storage is set up, a Reel’s cover picture is copied the same way.

Stats reads each Reel’s views, reach, likes, comment count, shares, saves and watch time, and the Page’s or account’s follower count (and, for Instagram, its number of posts). It does not read comment text, messages or other people’s data. Facebook’s and Instagram’s sign-ins return to the app through two small pages on this website. Each only passes the platform’s answer on to StreamSpark on your computer, stores nothing and loads no outside content. The one-time sign-in code is part of the page’s address, so Cloudflare can see it; it cannot be exchanged for access without your app’s secret, which stays on your computer. Meta handles data under its Privacy Policy.

Stats

Stats shows how the posts you made through StreamSpark are doing, using only the permissions above and each post’s own account. Posts are read about every 30 minutes for their first 30 days, then once a day; follower counts once a day; X as described above. Each platform has a daily limit on reads. Only numbers are kept (and, for TikTok, the video details above), in a database on your computer. They are never sent anywhere else and stay until you delete them.

Storage and security

Sign-in tokens and service keys are encrypted with Windows account-bound encryption and are never shown again after you enter them. Other files, such as footage, clips and databases, are protected by your computer’s own security, not by extra encryption. Clips you have not picked, edited or posted are removed by the clip timer (24 hours by default, set in Settings). Other local data has no automatic deletion date.

Delete your data

Everything StreamSpark keeps is on your computer or with the services you connected. To delete it:

For help, or to ask about data we may hold about you (for example, if you emailed us), email wonderlast888@gmail.com with “Delete my data”. We will confirm when it is done.

Questions and requests

Email wonderlast888@gmail.com with privacy questions or complaints, or to ask about access, correction or deletion. Do not send passwords or account tokens. We may need to confirm your identity. Platforms and services handle requests about their own records. If you email us, we use your address and message only to respond. Changes to this policy will be published here with an updated date.