Welcome to Kodezi Chronos-1
The first debugging-focused language model built for repository-scale understanding, multi-file reasoning, and autonomous code maintenance.

Kodezi Chronos-1 is the first debugging-first language model built for repository-scale understanding, multi-file reasoning, and autonomous code maintenance.
Unlike traditional code-generation models, Chronos-1 is trained specifically for repair, diagnosis, retrieval, and reasoning across entire repositories.
Chronos-1 is the intelligence layer powering Kodezi OS, CLI, and Web IDE.
Core Architecture Modules
These documents explain how Chronos-1 operates internally across retrieval, reasoning, validation, and memory systems.
Multi-Source Input Layer
How Chronos ingests logs, stack traces, code, tests, and repository history.
Adaptive Retrieval Engine (AGR)
Graph-based multi-hop retrieval that finds the exact files needed for a fix.
Debug-Oriented LLM Core
The transformer model fine-tuned for multi-file debugging and patch generation.
Autonomous Fix Loop
Runs the autonomous fix-test-refine loop with validation feedback.
Persistent Debug Memory (PDM)
Repository-specific memory that improves accuracy over time.
Execution Sandbox
Validates patches with real tests, linting, type-checking, and runtime checks.
Explainability Layer
Generates root-cause explanations, patch summaries, and documentation updates.
Getting Started
Availability Timeline
Q4 2025: Beta Access
Available to Kodezi Pro users (CLI + Web IDE)
- Repository-scale debugging
- Test-validated fixes
- PDM learning on your repo
Q1 2026: General Availability
Full access across platforms:
- Kodezi OS
- Kodezi CLI
- Kodezi Web IDE
- Kodezi API
Choose Your Platform
- Kodezi OS โ autonomous code maintenance
- Kodezi CLI โ terminal debugging
- Kodezi Web IDE โ browser-based debugging
- Kodezi API โ programmatic automation