Credentials are encrypted
Store keys, tokens and passwords are encrypted with AES-256-GCM before they are stored, decrypted only by the code that connects to your platform, and never shown again, logged or put in a URL.
Trust
Security and privacy
FeedManager holds read-only keys to your store and a copy of your catalog. Here is how that is protected, in plain language.
Passwords and keys are hashed
Account passwords use argon2id. API keys and one-time tokens are stored only as hashes, so a copy of the database does not hand anyone a working key.
Accounts are isolated
Every database query and every search is bound to one account. Products are stored per account, and files are served only through that feed's own address.
Access control
Manager and user roles, invitations by email, optional two-factor authentication — and mandatory two-factor for everyone on our side who can reach the platform. API keys carry scopes, so a key for your ERP can't be used by an AI agent, which needs the mcp scope.
Everything is logged
Changes to sources, feeds, rules, members and settings are recorded with who did them and when, and kept as long as your plan's history allows.
Untrusted data stays untrusted
Merchant files are parsed with entity resolution and network access off, with limits on size, depth and archive contents, so a hostile file cannot reach our systems. Store webhooks are signed (HMAC-SHA256) and only name a product — the product itself is read again through the connector. Pictures uploaded for Creative Studio are decoded safely and re-encoded, so only pixels — no scripts, camera or location data — reach the image server; SVG is refused.
Your URLs are fetched safely
Addresses you enter are resolved and checked before we connect: private, loopback and cloud-metadata addresses are refused, and the check is repeated after every redirect. Page Checks read only store domains you confirmed, within robots.txt, one page every three seconds.
Hosted in the EU
The platform runs on dedicated servers in Germany behind Cloudflare, with a firewall that only accepts traffic from the CDN and administrative access from known addresses.
Backups you can rely on
Nightly encrypted backups with hourly database logs, stored off-site, kept 30 days, and restored as a drill on a schedule — not only when something breaks.
Privacy
Your data stays yours
For the products, files and credentials you put into FeedManager, you are the controller and we are the processor. We process them to run the service you asked for — nothing else.
- A Data Processing Agreement is part of the Terms, with the processing details, our measures and our sub-processors listed.
- Export your account's data — settings, sources, feeds, rules, products and the activity log — as a downloadable archive.
- Close an account and it can be reopened for 30 days; after that everything is deleted, and gone from backups within another 30 days.
- We never sell your data, use it for advertising, or train AI models on it.
- Google access is chosen per connection — Analytics, Merchant Center, Search Console — and used only to read, with one exception you switch on per feed: instant price and stock updates in Merchant Center. It is revoked at Google when you disconnect.
- People you add as recipients of a scheduled report can take their address off the list from every report email, without a sign-in.
- The panel uses only the cookies it needs to work — no advertising or analytics cookies inside the product.
- Payments run through Stripe; card numbers are entered on Stripe's pages and never reach our servers.
GDPR and KVKK
FeedManager is built and run by a Turkish company serving customers in Europe, so both frameworks apply and we treat them as one standard.
- Access, export and deletion requests answered within 30 days.
- Retention periods written down for logs, runs, exports and backups.
- Sub-processor changes announced 30 days in advance.
- Breach procedure with notification deadlines, rehearsed as part of our runbooks.
We hold no ISO or SOC certification today, and we would rather say so than imply otherwise. If your procurement needs specific answers, ask us — we will fill in your questionnaire.
AI
AI with hard limits
AI features send product data to the AI provider that answers a run. They are built so that a feed full of untrusted text can't turn them against you.
How AI works in FeedManager- Models from Anthropic, OpenAI and Google. The panel names the company next to each model; your account can block any provider, and failover only uses the ones you allow.
- Product text is passed to the model as data, never as instructions, and every answer must match a fixed shape before it is used.
- AI can't write prices, identifiers, product links or availability — whatever a product's text says.
- Images are downloaded by us through the same checked fetcher as your files and sent to the model directly; no AI provider is given your image URLs to fetch.
- Suggestions are reviewed before use, stored next to your data and undone in one step. Your own provider keys are encrypted and shown only as their last four characters.
Reporting a vulnerability
If you believe you have found a security problem, write to [email protected] with enough detail to reproduce it. We will confirm we received your report, keep you updated while we fix it, and credit you if you would like that. Please give us reasonable time before publishing, and don't access other people's data or disrupt the service while testing.
Machine-readable contact details: /.well-known/security.txt
Questions from your security team?
Send them over — we answer questionnaires and explain how anything on this page works in practice.
Private beta — accounts by invitation.