Decisions
Why was this architecture chosen?
Contorium
Project Intelligence Layer
for AI-assisted development.
AI can read your code.
Contorium helps it understand why your project became what it is.
Preserve project decisions, reasoning, constraints, history, and evolving state — across AI tools, models, and sessions.
Local-first · Open Source · MCP-compatible
Scroll ↓Git commit
Decision
Reason
Impact
AI understands
AI can generate code.
But it doesn't know the history behind it.
A codebase is more than its current files. It also contains:
Reasoning
What trade-offs led here?
Constraints
What must not be changed — and why?
Evolution
How did the project become what it is today?
Current State
What is happening now, and what comes next?
AI coding tools can inspect the repository.
But the repository alone doesn't contain all of its intelligence.
Give your project a persistent intelligence layer.
Contorium preserves the knowledge that normally disappears between AI sessions — not just chat history, not just prompts, not just a snapshot of your codebase.
Contorium captures the project's:
State
Intent
Decisions
Reasoning
Outcomes
Evolution
So AI can continue from what the project already knows — instead of reconstructing everything from scratch.
One project. One intelligence layer. Every AI tool.
Your Project
Contorium
Project Intelligence Layer
Cursor
Claude Code
Codex
Same Understanding
Ask your project.
Instead of asking an AI to reconstruct your project from scratch:
Why was MCP added?
Contorium can trace the decision, alternatives, and reasoning behind it.
What changed this week?
See the project's evolution instead of reading through commits and chats.
What do we know about authentication?
Connect decisions, modules, constraints, and history around one topic.
What should I know before changing this?
Surface the project knowledge that matters before the next change.
You
Why was MCP added?
Contorium
Decision #024
VALID
MCP was introduced because:
- Multiple AI tools needed shared context
- Vendor lock-in was avoided
How it works
Contorium separates project intelligence from the AI tool using a local, structured architecture.
AI Tools
CIL
Cognitive Interaction Layer
Project Intelligence
.contora/
CIL — Cognitive Interaction Layer
The interaction layer that lets humans and AI agents query, capture, and transfer project intelligence.
Project Intelligence
Decisions, reasoning, history, relationships, constraints, and evolution — structured and local.
.contora/
The intelligence stays with the project — locally, inside the repository.
Project intelligence, made queryable.
Project History
Understand how your project evolved over time.
Decision Center
Preserve decisions, alternatives, trade-offs, and reasoning.
Knowledge Graph
Connect modules, decisions, people, constraints, and events.
Time Travel
Reconstruct what the project knew at a specific point in time.
Cognitive Health
Detect missing reasoning, stale decisions, and conflicting knowledge.
Project Story
Turn project history into an understandable narrative for humans and agents.
Project DNA
Create a compact identity of the project for AI handoffs.
Evolution Journey
See how the project changed — and why.
Your tools change. Your project doesn't.
Use the AI coding tool you prefer. Switch tools without losing the project's accumulated understanding.
Cursor
Claude Code
Codex
Gemini
VS Code
MCP
One project.
One intelligence layer.
Every AI runtime.
Not another AI memory system.
Chat Memory
Remembers conversations.
RAG
Retrieves relevant information.
Git
Tracks code changes.
Contorium
Preserves the intelligence behind the project.
What changed → Why it changed → What happened → What remains true → What comes next
Your project shouldn't forget
what it has learned.
Contorium gives AI coding tools a persistent intelligence layer — so every session can build on what came before.
Local-first · No telemetry · Open Source · MIT License