Security Overview
Security and privacy are foundational design constraints in F7 — not afterthoughts. This page summarizes our security posture for CISOs, procurement teams, and security evaluators.
Core Security Posture
- Defense in depth. Six distinct security layers protect data from endpoint to storage.
- Privacy by design. The agent processes data locally before transmitting only structured, PII-scrubbed metadata. Content is never captured.
- Zero-trust enrollment. Agents authenticate via a cryptographic enrollment flow that produces per-device credentials stored in the operating system's secure credential store.
- Modern cryptography. All algorithms are current-generation with no legacy primitives.
- Memory-safe codebase. The platform is built in Rust, eliminating entire classes of vulnerabilities (buffer overflows, use-after-free, data races) at compile time.
Six Layers of Defense
| Layer | What It Does |
|---|---|
| 1. Transport Security | TLS 1.3 for all communications. HSTS enforced. Certificate pinning for agent connections. |
| 2. Authentication | EdDSA-signed agent JWTs with per-device credentials. Argon2id passwords or OAuth 2.0/OIDC SSO for admins. IdP directory sync with auto-provisioning. |
| 3. Authorization | Hybrid ReBAC+ABAC policy decision point. Four roles with manager-chain scoping, purpose-specific enforcement, app-category delegation, and k-anonymity. |
| 4. Database Isolation | Row-Level Security on every org-scoped table with read and write enforcement. |
| 5. Audit & Monitoring | Immutable, trigger-protected audit log. Cannot be modified or deleted even by the application. |
| 6. Encryption at Rest | AES-256-GCM for server-side secrets. AES-256 encrypted local database on each device. |
Compliance Quick Reference
| Question | Answer |
|---|---|
| Data encrypted in transit? | Yes — TLS 1.3 |
| Data encrypted at rest? | Yes — AES-256-GCM (server), AES-256 (agent) |
| Role-based access control? | Yes — Hybrid ReBAC+ABAC with PDP, four roles, manager-chain scoping, app-category delegation |
| SSO / OIDC? | Yes — Entra ID, Okta, Google Workspace, JumpCloud, generic OIDC |
| Tenant isolation? | Yes — PostgreSQL Row-Level Security on all org-scoped tables |
| Audit logging? | Yes — Immutable (trigger-protected), 24-month retention |
| Modern password hashing? | Yes — Argon2id |
| Memory-safe language? | Yes — Rust |
| Dependency auditing? | Yes — Trust Center deploy CI audits all of its dependencies before publishing; product dependency evidence is release-scoped |
| Right to erasure? | Yes — Full data deletion on request |
| Can employees pause collection? | On request — an administrator pauses observation, keeps the app exclusion list and sets work hours |
Audit & Testing
F7 undergoes regular internal security audits following a structured three-pass methodology. This Trust Center's deploy workflow audits all of its documentation dependencies for known vulnerabilities at moderate severity or above before publishing, and fails on any such advisory that has no dated, reasoned exception; product dependency evidence is tracked in release evidence for the owning private repositories.
Independent third-party penetration testing and SOC 2 Type II certification are on our roadmap.
Deeper Dives
- Architecture — The six-layer model in detail
- Authorization — PDP authorization model, obligations, and IdP integration
- Encryption — Complete cryptographic inventory
- Agent Security — How data is protected on the device