Release: v16.0.0-rc1
Security Standard: Zero-Trust Architecture & Cryptographic Integrity

• Password Hashing: Cryptographic password storage using Argon2id (passlib.handlers.argon2 with $m=65536, t=3, p=4$), designed to increase resistance to GPU/ASIC-assisted password cracking.
• JWT Asymmetric Tokens: API requests are authenticated using RS256 asymmetric signing (RSA 2048-bit keys) with token expiration and revocation tracking.
• Role-Based Access Control (RBAC): Strict permission boundaries separating candidate, recruiter, faculty, mentor, and admin roles.

• Tenant Boundaries: All organizational data (e.g. employee skills, candidates, jobs, analytics) is isolated by organization_id derived exclusively from authenticated user token claims.
• IDOR Protection: Direct database queries filter strictly on user_id == current_user.id or organizational scope; direct query parameter overrides are forbidden.

• HMAC-SHA256 Signatures: Issued digital credentials compute HMAC-SHA256 authenticated integrity protection over a canonical, normalized payload: $$\text{Payload} = \text{cred_id} ,|, \text{user_ref} ,|, \text{competency} ,|, \text{level} ,|, \text{issuer} ,|, \text{issued_at} ,|, \text{version}$$
• Verification Engine: Altered skill names, scores, or dates cause immediate verification failure without silent degradation.

• High-Impact Action Guard: Autonomous workflows are governed by Autonomy Level 3 policies. High-impact operations (APPLY_JOB, MESSAGE_RECRUITER, PUBLISH_CREDENTIAL, CHANGE_GOAL) require explicit user confirmation (is_confirmed=True) before execution.
• Unilateral Action Ban: AI agents are strictly forbidden from independently hiring, rejecting, or submitting irreversible applications without candidate or recruiter sign-off.