What if it didn't have to be? Watch one conversation flow across six platforms. Nothing lost. Nothing re-explained.
Six platforms. One memory. Nothing lost.
Kenotic Continuity Bridge is a local memory layer that runs on your device. Connect any AI that speaks MCP: Claude, GPT, Gemini, Cursor, local models through llama.cpp or Ollama. One memory file. Every platform. You own it.
{
"mcpServers": {
"kenotic": {
"command": "py",
"args": ["-m", "mcp.server"],
"env": { "KENOTIC_DB_PATH": "~/.kenotic/memory.db" }
}
}
}Tested against ATANT, an open evaluation standard for AI continuity (arXiv:2604.06710). Corrections tracked. Temporal ordering preserved. Disambiguation across entities. No LLM at read time. Retrieval is deterministic, so answers don't change when you switch models or when the provider ships an update.
Each AI stores its own flat list of facts about you
One structured memory file tracks facts, corrections, temporal ordering, and context
Switch platforms = start over with that platform's separate memory
Switch platforms = same memory, same understanding, nothing lost
Memory lives on their servers
Memory lives on your device. A SQLite file you can inspect, export, or delete
Stores isolated key-value pairs like "likes Python" and "lives in Portland" with no relationship between them
Stores structured traces: identity, corrections, temporal ordering. The system reconstructs context, not isolated facts
No cloud. No upload. The memory file is on your machine.
Claude today, GPT tomorrow, a local model when you want privacy. Context follows you.
When facts were corrected, when context shifted, what superseded what.
It's a SQLite file. Open it, inspect it, back it up, delete it. No API required.
Four human stories. Sessions across Claude, ChatGPT, Gemini, and local models. Corrections, platform switches, emotional arcs. See what continuity looks like when memory persists.