Security & Data Protection Policy
Security & Data Protection Policy
Last Updated: November 15, 2026
1. Overview
CostRadar.ai is operated by Radrly spółka z ograniczoną odpowiedzialnością (KRS 0001215510). As of October 9, 2026 the company has one person with production access: Marcin Nowak, CEO, who is also the Information Security Owner and the incident lead. There are no employees and no contractors with production access. This page describes controls that are in operation, plus items that are planned and dated. It implements Information Security Policy RADRLY-ISP-001 v1.0 (effective October 9, 2026), which is reviewed at least every 12 months. The next scheduled review is October 9, 2027.
Our security practices are designed to support GDPR/CCPA-oriented privacy controls and platform requirements (including Shopify App Store security expectations for Shopify installs). CostRadar is not SOC 2, ISO 27001, or PCI DSS Level 1 certified, and we have not commissioned a third-party penetration test.
2. Security Principles
Our security framework is built on the following core principles:
Confidentiality
Ensuring data is accessible only to authorized parties through encryption and access controls
Integrity
Protecting data from unauthorized modification or corruption
Availability
Maintaining reliable access to services and data when needed
Accountability
Logging and monitoring all access and changes to maintain auditability
3. Data Encryption
3.1 Encryption in Transit
All data transmitted between your browser and our servers is protected using:
TLS 1.2 and TLS 1.3. TLS 1.0 and TLS 1.1 are disabled. We do not restrict traffic to TLS 1.3 only.Certificates from Let's Encrypt, renewed automaticallyHSTS (max-age=31536000; includeSubDomains) is sent on costradar.ai. It is not yet sent on supabase.costradar.ai.3.2 Encryption at Rest
We protect stored credentials and sensitive integration secrets as follows:
Platform OAuth tokens are encrypted with AES-256-GCM before they are stored. If the encryption key is missing, the server refuses to store tokens.Database volume encryption is not shown in this repository.An encrypted backup script (AES-256 via gpg or age, optional off-server copy) is in the repository. A daily encrypted off-server job is installed on the production host.Encryption keys and app secrets live in the server environment and in GitHub Actions encrypted secrets.3.3 Password Security
Passwords are hashed using bcrypt with individual saltsPasswords are never stored in plain text or reversible encryptionMinimum password complexity requirements enforcedSupport for password reset via secure email verification4. Authentication and Access Control
4.1 User Authentication
JWT (JSON Web Token) based authenticationSecure session management with automatic timeoutProtection against common session abuse patternsRate limiting and abuse controls on sensitive auth endpoints where implementedEmail verification / OAuth sign-in flows as configured in the product4.2 OAuth Integration Security
For third-party integrations (Shopify, Amazon, WooCommerce, Meta Ads, Google Ads, TikTok Ads, Klaviyo, and other platforms you connect):
OAuth 2.0 authorization flow with state parameter for CSRF protection (where OAuth applies)Platform tokens are encrypted at rest with AES-256-GCM. The server will not store a token if the encryption key is missingTokens scoped to minimum necessary permissions (read-only where possible)Token refresh/validation according to each providerDisconnect deletes stored access and refresh tokens for TikTok Shop, BigCommerce, Allegro, and eBay, and clears the Shopify token on app uninstall. TikTok Shop does not receive a remote revoke call from CostRadar; the seller can also remove the app in TikTok Shop. TikTok Shop order and fee rows are deleted within 30 days of disconnect by the platform orders job.4.2.1 Platform Integrations
Additional security measures for advertising and marketing platform integrations:
Data Access Scope: Read-only access to ad spend metrics only (no customer data, ad creative, or targeting). Google Analytics 4 is separate: read-only analytics reports for properties you connect.Supported Platforms: Meta Ads (Facebook/Instagram), Google Ads, Google Analytics 4, TikTok Ads, Pinterest Ads, Snapchat Ads, Amazon AdsOAuth Permissions: Meta Ads: ads_read (read-only ad spend data)Google Ads: https://www.googleapis.com/auth/adwords (used read-only)Google Analytics 4: https://www.googleapis.com/auth/analytics.readonly (read-only)TikTok Ads: ad_account.readonlyOther platforms: Read-only equivalentsData Collected: Campaign names, daily ad spend amounts, account names, currencies (NO customer lists, emails, phone numbers, or ad creative). Google Analytics 4: sessions, users, and conversions from reports you authorize.Token Management: Encrypted at rest, automatic expiration notifications, secure refresh flowUser Control: Users can disconnect anytime, historical data retained (per DPA), no future syncs4.3 Role-Based Access Control (RBAC)
Merchants can access only their own data. PostgreSQL row-level security and per-request checks enforce that.Separation of duties is not possible with one operator. Compensating controls are encrypted secrets, deploys from GitHub Actions, and an annual self-review.The database service-role key is used only on the server. It is not shipped to the browser.5. Infrastructure Security
5.1 Cloud Infrastructure
CostRadar production architecture:
Self-hosted Supabase-compatible stack
API/DB at supabase.costradar.ai (CostRadar-operated; not Supabase Inc. hosted US as primary)Frontend delivered via Netlify and VPS static hostingPostgreSQL with application-level authorization checksSecurity patching and operational hardening on our infrastructureCloudflare Turnstile for signup bot protection5.2 Network Security
Nginx is the internet-facing proxy for costradar.ai and supabase.costradar.ai. There is no Cloudflare WAF in front of the API. Cloudflare is used for Turnstile on signup only.The API host rate-limits at 30 requests/second per IP with a burst of 60.Password login lockout: 10 failed attempts lock the email and IP pair for 30 minutes.The information security policy requires a host firewall (public ports limited to 80, 443, and SSH), SSH key-only login, fail2ban or equivalent, and Ubuntu unattended-upgrades. Those host settings are not stored in this repository and must be confirmed on the server.5.3 Application Security
Input validation and sanitization to prevent injection attacksProtection against XSS (Cross-Site Scripting)CSRF (Cross-Site Request Forgery) token protectionContent Security Policy (CSP) headersGitHub Dependabot is configured for npm, GitHub Actions, and the Deno functions that ship a deno.json. CI runs npm audit and a working-tree secret scan. The scan reports findings and does not fail the deploy workflows.Changes reach production through GitHub Actions on the main branch.6. Data Backup and Recovery
6.1 Backup Strategy
A daily encrypted off-server backup job is installed on the production host. The script encrypts a pg_dump with AES-256 (gpg or age) and copies it off the server.Local backup folders created by the scripts are pruned after 30 days. There is no point-in-time recovery and no geographic redundancy of backups.6.2 Disaster Recovery
Operator notes record RPO and RTO targets of 24 hours. We do not claim a 4-hour recovery time.A restore test to a non-production database is planned at least every 6 months. No restore-test record exists yet.7. Monitoring and Logging
7.1 Security Monitoring
Application and infrastructure logging for security-relevant eventsMonitoring of failed authentication and abuse patterns where instrumentedAPI rate limiting and abuse detection where implementedPromotional code usage monitoring and fraud preventionAlerting to the on-call/security contact (support@radrly.com) for high-severity events7.2 Audit Logging
Logging of authentication and significant account eventsIntegration connection and disconnection eventsPromotional code redemption and usage eventsLog retention: typically up to 90 days7.3 Incident Detection
Operational alerts for suspicious or elevated error patternsManual review and escalation via our incident response process8. Third-Party Security
8.1 Sub-processor Security Assessment
All third-party service providers are carefully vetted:
Security documentation review for critical vendors (SOC 2/ISO where the vendor publishes it)Data Processing Agreements / DPAs where requiredPeriodic review of sub-processor list in our Trust Center / DPAAlignment with GDPR transfer requirements for US vendors (e.g., SCCs)8.2 Current Sub-processors
Aligned with Privacy Policy / DPA. Full detail and transfer terms: DPA §5.
Self-hosted infrastructure + Netlify
API/DB at supabase.costradar.ai (not Supabase Inc. hosted US as primary); frontend via Netlify + VPS
Cloudflare
Turnstile bot protection on signup and similar flows
OpenAI, DPA / transfer terms as applicable
AI Insights, Chat, Briefing, Support Chat
Stripe / Shopify Billing
Payment processing (vendor PCI scope; not CostRadar PCI Level 1)
Shopify / Amazon / WooCommerce / BigCommerce
Store platform access when Customer connects them (OAuth / SP-API / store API)
Optional ads / email / Zapier
Meta, Google, TikTok Ads, Pinterest, Snapchat, Klaviyo, Zapier, only when Customer connects them
GTM / GA / Google Ads
Marketing-site measurement subject to Cookie Consent
Google & GitHub
OAuth identity providers for sign-in
SMTP email provider
Transactional and support email delivery
9. Employee Security
9.1 Who has access
One person, Marcin Nowak (CEO), holds production, database, and vendor admin access.The information security policy requires MFA on GitHub, the hosting account, DNS, email, Stripe, and platform partner portals including TikTok Shop Partner Center. Confirm those settings in each provider before treating MFA as evidenced.There is no separate CTO, DBA, support staff, or developer role.Access to production accounts is reviewed at least every 12 months. No recorded review exists yet.9.2 Training
The Information Security Owner completes security and privacy awareness refresh training at least yearly.9.3 Future staff
Any future contractor gets the minimum access needed, their own account, and a confidentiality agreement before access starts. Access is revoked the day the engagement ends.10. Vulnerability Management
10.1 Vulnerability Scanning
GitHub Dependabot and npm audit, as configured in this repository.Ubuntu security notices and unattended-upgrades are required on the host. Confirm that the package is enabled on the server.An external TLS and header check at least quarterly.We have not performed third-party penetration testing. We will commission it when revenue and partner requirements justify it, and we will not claim it until a report exists.10.2 Patch Management
Critical: 72 hours. High: 14 days. Medium: 30 days. Low: 90 days or the next release. Measured from when we become aware of the finding.Findings are tracked as GitHub issues labelled security.A monthly patch window covers OS packages, container images, and dependencies.10.3 Responsible Disclosure
We encourage security researchers to report vulnerabilities responsibly. Contact support@radrly.com with details. We commit to:
Acknowledging reports within 48 hoursProviding status updates during investigationCrediting researchers (with permission) after resolutionNot pursuing legal action against good-faith researchers11. Compliance and Certifications
CostRadar security and privacy program is oriented around:
GDPR (General Data Protection Regulation) principles and customer rights toolingCCPA (California Consumer Privacy Act) where applicableShopify App Store security / GDPR webhook requirements for Shopify installsAmazon Selling Partner API Data Protection Policy (WAF, credentials encrypted at rest, 30-day deletion, incident point of contact)OWASP Top 10 security best practicesPayment card data is handled by Stripe and/or Shopify Billing under those vendors' PCI programs. CostRadar does not claim PCI DSS Level 1 certification for itself, and does not claim to be SOC 2 certified. Some vendors we use may publish their own SOC 2 / ISO reports.
12. Incident Response
In the event of a security incident:
Immediate containment and investigationNotification to affected users within 72 hours (GDPR requirement)Coordination with data protection authorities as requiredPost-incident review and remediationImplementation of preventive measuresFor detailed incident response procedures, see our Security Incident Response Policy.
13. Security Reviews and Updates
We continuously improve our security posture through:
The information security policy is reviewed at least annually, and after a significant incident or a new platform integration. It is not reviewed quarterly.No annual third-party security audit has been completed.Responsible disclosure: support@radrly.com14. Contact Information
For security-related inquiries or to report a vulnerability:
15. Policy Updates
This Security Policy is reviewed and updated regularly to reflect changes in our security practices and compliance requirements. Material changes will be communicated via email and posted on this page with an updated "Last Updated" date.