RCT Knowledge Vault Genome
The core knowledge base of RCT Ecosystem — a Living Knowledge System powering all platforms.
1,068+
Total Files (Snapshot v1.0)
3
Separate Vaults
100%
User-Key Encrypted
3-Vault Structure
Vault-1
DelentiaAI Vault
The primary agent knowledge vault. Contains ARTENT operating protocols, intent classification training data, WF-00 through WF-05 phase definitions, error taxonomy, and the full history of agent session data. Organized by intent category and domain.
Vault-2
SignedAI Vault
The verification knowledge vault. In the design it holds scoring calibration data, model comparison benchmarks, jury configurations and attestation templates. Only the jury configuration file exists today; no hallucination-pattern library or calibration data has been built.
Vault-3
Middleware Vault
The integration and protocol vault. Contains JITNA RFC-001 specification files, API gateway configurations, RCTDB schema definitions, LLM adapter specifications, and system health monitoring baselines.
Data Sovereignty Principles
User Ownership
All user-contributed data in the RCT Knowledge Vault is owned by the user who created it. Delentia Labs holds no license to use it for training or analytics without explicit M-primitive consent declaration.
Selective Disclosure
Users can share specific Vault segments with agents, collaborators, or external systems without exposing their entire vault. Disclosure is per-file, per-session, or per-role — never all-or-nothing.
Encryption at Rest
All RCT Knowledge Vault data is encrypted with user-held keys. RCT infrastructure operators cannot read vault contents. The encryption key is never transmitted — only used locally or in secure enclaves.
TTL Declarations
Every memory write includes an explicit TTL (Time-to-Live) from the M-primitive. No data persists forever by default. Session memory defaults to 24h. Project memory requires explicit declaration. No silent accumulation.