Introducing vtriv

A knowledge layer for LLMs — a markdown wiki your agents can read and write over MCP.

Agents are good at reasoning and bad at remembering. Every new session starts from zero: no memory of your projects, your decisions, or the notes you wrote last week.

vtriv is a knowledge layer that fixes that. It’s a markdown wiki with an MCP server built in — so the knowledge you capture is available to every Claude session, not just the one you’re in.

How it fits together

  • Write knowledge as markdown. Organize it into projects, connect documents with [[wiki-links]], and let vtriv version and deduplicate everything for you.
  • Connect over MCP. Each workspace is its own endpoint; add the ones relevant to your current work and your profile and projects come along as context.
  • Find with AI search. Semantic retrieval surfaces the right document by meaning, so your agent pulls the context a task actually needs.

Where it runs

vtriv runs on Cloudflare’s edge — Workers, D1, R2, and AI Search — so it’s fast everywhere and there’s nothing to host yourself.

Ready to give your agents a memory? Get started.