0 leaks
of tokens in the entire platform history
Trust Center • Zero-Trust architecture
Enterprise-grade security
Isolated session containers, hardware-backed AES-256-GCM token vault, personal-data masking and a full audit trail. We never ask for passwords — official Meta* OAuth 2.0 only.
SOC 2 Type II (in progress)GDPR152-FZ (for RU)Meta* App ReviewISO 27001 (roadmap)
Zero-Trust Session SandboxingLayer active
Click a layer to see how it protects your accounts
Each account session lives in its own sandbox container: a failure or compromise of one never touches the others.
AES-256-GCM
encryption + HSM keys, rotation every 90 days
365 days
immutable audit-trail retention
24/7
anomaly monitoring and token auto-revoke
Defense in depth
Four layers between threats and your account
Compromising one layer gives an attacker nothing but encrypted garbage.
Zero-Password model
No passwords, no browser cookies. Official Meta* OAuth 2.0 only, with minimal scopes and one-click revoke.
Session sandboxes
Kernel-level isolation: network, files and memory of each session are separated. Egress only to allowlisted Meta* and Telegram hosts.
PII minimization
We collect only what the product needs. Masking in UI, logs and notifications. Full deletion on request within 72 hours.
Team access control
RBAC roles, SSO, mandatory 2FA and a per-employee action log. Least privilege by default.
Backup & recovery
Encrypted backups every 6 hours in two EU regions. RPO 6 hours, RTO 4 hours, quarterly restore drills.
Supply-chain security
SBOM for every build, dependency scanning in CI, signed releases. No grey proxies or browser emulation.
Who needs this protection
Three scenarios where a mistake costs an audience.
25+ accounts
Agencies with brand portfolios
Isolated client workspaces: one brand’s tokens are invisible to another brand’s team. Client Portal with separate auth.
€20M+ fines
Companies under GDPR
EU data residency, DPA on request, deletion on demand. For Russian clients — processing under 152-FZ.
1 click
Brands with ad budgets
An ad-account compromise is stopped by token auto-revoke in seconds. Dark Post Shield keeps hiding spam.
Engineering Deep Dive
How it works under the hood
Five engineering decisions verified by our external audit.
1. Token encryption
AES-256-GCM, a nonce per record, keys in HSM with no export. Decryption happens only inside the vault process — tokens never reach logs.
2. BYOS & BYOK
Bring Your Own Secret/Keys: large clients attach their own HSM or KMS. Revoking the key on your side instantly invalidates all tokens.
3. Meta* App Review
We pass the official Meta* App Review justifying every permission. Only approved scopes and Graph API v20.0 — no scraping.
4. Token rotation
Auto-renewal of long-lived tokens 7 days before expiry, encryption-key rotation every 90 days, instant revoke on login anomaly.
5. Audit trail
Append-only log: publish, approve, comment-hide, login. Every event signed, SIEM export (Syslog/HTTPS), 365-day storage.
Incident Response: 4 steps in 1 hour
A response plan we rehearse quarterly.
- 0–15 min: contain: Auto-revoke affected tokens, freeze sessions, alert the on-call engineer and the account owner in Telegram.
- 15–30 min: assess: Audit-trail analysis: what was done, what data was touched. Severity classification P1–P4.
- 30–45 min: remediate: Key rotation, token reissue via OAuth, blocking the vector (IP, key, employee).
- 45–60 min: notify: Client report: what happened, what was affected, what was fixed. For GDPR — DPA notification within 72 hours.
Threat → protection matrix
What can go wrong — and which layer stops it.
| Threat | Boostly protection | Outcome |
|---|---|---|
| Employee password phishing | Zero-Password + SSO + 2FA | No password exists — nothing to steal |
| Token database leak | AES-256-GCM + no-export HSM | Ciphertext without keys is useless |
| Single session breach | Container isolation + egress allowlist | Attack never leaves the sandbox |
| Insider abuse | RBAC + audit trail + PII masking | Every action is signed and visible |
| Ad-account compromise | Token auto-revoke + 24/7 alerts | Access closed in seconds |
Security you can verify
Request the security whitepaper, a pentest report under NDA, or trial access with an isolated workspace.
Security you can verifyIsolated workspace • One-click access revoke
Layer activeStart 3 days free
Security FAQ
No. We run a Zero-Password model: authorization only via official Meta* OAuth 2.0. No password exists in the database or logs — there is nothing to steal.
The attacker gets AES-256-GCM ciphertext. Keys live in a hardware HSM with no export, so tokens cannot be decrypted outside our vault.
Yes. For Russian clients data is processed under 152-FZ: minimal scope, deletion on request within 72 hours, data-processing agreement.
Automatically in seconds on anomaly (new IP/geo, mass actions). Manually — one “Revoke all tokens” button in the dashboard and a bot command in Telegram.
Nobody without need: RBAC, PII masking in the UI, every data access written to the audit trail. Access granted per ticket with TTL.
Yes: annual third-party pentest, dependency scanning on every CI run, SOC 2 Type II in progress. Reports available under NDA.