The knowledge layer for LLMs
Durable context
for every session.
vtriv is the persistent context layer that makes your AI assistants actually useful across sessions and tools. One MCP connection gives any client durable memory and structured context it can read and write.
One layer your LLMs read and write.
Connect once over MCP and your whole layer is in scope — for Claude or any MCP client.
Who you are, your stack, your preferences — always in scope. Stop re-explaining yourself to every new chat.
Reusable Agent Skills your LLM can load on demand. Package expertise once; use it in every session.
A markdown filesystem the LLM organizes freely by path. It reads, writes, and structures it however the work demands.
Bodies of work with tasks, notes, and a timeline — the durable context behind everything you're shipping.
How it works.
Set up your profile, add skills, and organize knowledge as markdown. Spin up projects with tasks, notes, and a timeline — in the web UI or straight from your LLM.
Add a single global MCP connection to Claude, Claude Code, or any MCP client. Your whole layer — profile, skills, knowledge, projects — comes along automatically.
Hybrid semantic search surfaces the right context the moment a task needs it. New knowledge your assistant produces lands back in the layer for next time.
Hybrid semantic search across everything.
Profile, skills, knowledge, and projects are all searchable by meaning, not just keywords. Your assistant asks a question; the right context comes back — no matter where you filed it.
Give your LLMs a memory.
Free to start. Stop re-explaining yourself to every new chat — build a context layer your assistants read and write.
Get Started