⚡ Unified Data Layer & Multi-Model Engine

The Multi-Model Engine
Modern Data Runs On.

Write once in Colt. Execute across PostgreSQL, Redis, Mongo, and Firestore with physical tenant isolation, zero parameter leakage, and automatic capability checks.

⚡ Launch Console 🔑 Sign In / Register 📖 Documentation
Compiler Active · Lexer & Parser Verified
PostgreSQL · Redis · Firestore · Mongo
TenantGuard Isolation Active
Engine Architecture

Engineered for absolute speed. Built without limits.

A high-performance developer console with first-class query compilation, interactive execution plans, and strict security boundaries.

⚡ Declarative Engine STAGE 01 — 04

Unified Query Planner

One declarative syntax compiles down to optimized PostgreSQL parameters, Redis KV commands, and Firestore paths. Zero injection attacks guaranteed.

DEFINE COLLECTION books STRICT @backend(neon) {
  id: uuid @key,
  title: text,
  embedding: vector(1536) @index(hnsw)
}
FIND books SELECT title, id ORDER BY views DESC LIMIT 10
🔒 Hardware Isolation

TenantGuard

Physical schema separation via ct_{tid} and runtime SET LOCAL ROLE ensures zero multi-tenant data leaks.

✓ Cryptographic Isolation Active
Cross-tenant query attempts are halted at the compiler boundary.
⚡ Upstash Redis

Sub-Millisecond KV

Prefix isolation with t:{tid}: keys for rapid session state, token caches, and rate-limiting.

🔥 Firebase Fullscale

Firestore & Realtime

Integrated with project coltanedb. Google OAuth, email authentication, and multi-tenant document collections ready for production.

🧠 Vector & HNSW

AI Agent Memory

Native vector types with HNSW index planning tailored for LLM context graphs and semantic recall.

🛡️ Strict Spillover Boundary D8 STRICT ENFORCED

Within-Family Boundaries — Zero Cross-Model Spillover

Relational tables fail over only to relational stores (Neon ↔ Supabase). KV pairs fail over strictly to KV stores. Document graphs never spill into SQL tables.

STAGE 01 Lexical Scan Splits source into typed tokens and diagnostics with COLT1xxx codes.
STAGE 02 Recursive AST Builds strongly typed AST of statements, types, and engine annotations.
STAGE 03 Semantic Guard Validates collections, field presence, and target backend placement.
STAGE 04 Plan Emission Generates parameterized SQL plans and route bindings for backends.
Frequently Asked Questions

Everything about ColtaneDB

What is ColtaneDB and the Colt language?

ColtaneDB is an open-source, web-first data platform built around Colt, a custom query language. The product is a language, compiler, control plane, and query router over existing databases (Neon, Redis, Mongo, Firestore), rather than a proprietary storage engine.

Do I need to install a CLI or IDE to use it?

No installation is required. Colt Studio is the Official Web-Only IDE for Colt. You can write, parse, plan, import, and export .colt files directly from this web console. Python developers can also use our official coltanedb package on PyPI.

How are credentials and multi-tenant data protected?

TenantGuard isolates every tenant into their own schema (e.g., ct_tnt_...) or path namespace. Credentials live exclusively in server-side encrypted stores or environment variables, and are never logged or exposed in client bundles.

Colt Studio |
query.colt Ln 1, Col 1
1
Visual Schema Designer

Model data visually without writing code

Add collections, configure types, establish relations, and export clean .colt declarations instantly.


        

      

    


    
    

      

        

          COLLECTIONS
          4 TOTAL
        

        

          
books

          
authors

          
users

          
metrics

        

      


      

        

          

            books
            TENANTGUARD
              ACTIVE
          

          

            
            
          

        


        

          
        

      

    


    
    

      

      

        

          
1. Platform Overview
          

          

            ColtaneDB provides a unified control plane and declarative language (Colt / .colt) over
            existing database providers. Rather than hosting a proprietary storage engine, ColtaneDB lexes, parses,
            validates, and emits parameterized execution plans targeted at PostgreSQL (Neon, Supabase), Redis (Upstash),
            Document stores (Firestore, MongoDB), and Edge SQLite (Turso).
          

        


        

          
2. Colt Query Grammar
          

          

            Colt syntax enforces strict typing, declarative backend placement via annotations
            (@backend(neon)), and clear primary key specifications (@key,
            @unique).
          

          
DEFINE COLLECTION users STRICT @backend(neon) {
  id: uuid @key,
  email: text @unique,
  tier: text @default("free")
}

FIND users SELECT email, tier WHERE tier == "pro"

        


        

          
3. TenantGuard
            Architecture

          

            Every tenant is designated an opaque ID (tnt_...). On PostgreSQL backends, TenantGuard creates
            dedicated physical schemas (ct_) and runs every query under SET LOCAL ROLE to
            guarantee that cross-tenant data leaks are physically blocked at the database kernel level.
          

        

      

    


    
    

    
    

      
404

      
Route Not Found

      

        The requested path does not match any known ColtaneDB console view or documentation route.