Git is incredible.
It remembers every file.
Every commit.
Every branch.
Every merge.
But Git was never designed to remember why.
⸻
Months later, developers ask questions Git cannot answer.
Why was this microservice split?
Why didn’t we use Kafka?
Why does this API still exist?
Why was this refactor abandoned halfway?
The information existed once.
Usually inside someone’s brain.
Or an AI conversation.
Or a meeting.
Then it disappeared.
⸻
As AI becomes part of software development, this problem becomes even bigger.
We’re generating more code than ever.
But we’re preserving less engineering knowledge than ever.
The faster code is written,
the faster reasoning disappears.
⸻
Imagine pairing Git with a second layer.
Git tracks files.
Another layer tracks decisions.
Git records what changed.
A cognitive layer records why it changed.
Together they create something much closer to how humans actually understand software.
Not files.
Not folders.
Not commits.
But intent.
That feels like a much stronger foundation for AI-native development than simply giving models larger context windows.
ContoriumLabs
/
contorium
Project Intelligence Runtime powered by CIL. Ask your project what happened, why it changed, and what comes next across any AI coding tool.
Contorium
Project Intelligence Runtime for AI Coding Tools
Contorium is a local-first system that preserves and reconstructs software project intelligence across AI coding environments.
It works with:
- Cursor
- Claude Code
- Codex
- Gemini CLI
- VS Code
- Any MCP-compatible runtime
Instead of repeatedly re-explaining architecture, decisions, and project state in every session, Contorium maintains a shared Project Intelligence Layer that all tools can access.
Core Idea
Git stores code history
Contorium stores project intelligence history
It captures not just what changed, but:
- Why it changed
- What decisions were made
- What the project is trying to become
- How it evolved over time
CIL — Cognitive Interaction Layer (v3)
CIL is the user-facing intelligence system.
It does not execute tasks.
It answers questions.
User Query
↓
Query Router
↓
Cognitive Kernel
↓
Engines (Event · Decision · State · Graph)
↓
Formatter
↓
Response
What CIL Can Answer
| Question | System |
|---|---|
| What happened? |
Top comments (0)