Introduction

Vaner is a local-first predictive context engine for AI coding workflows.

Think of it like a chess engine for context: during idle time, Vaner predicts likely next prompts, prepares useful context branches, and then serves the best package quickly when the real prompt arrives.

What Vaner does

• Runs a background ponder loop to pre-build likely context.
• Runs a prompt-time answer loop to rank and assemble the best context package.
• Stays model-agnostic, so you can use local or cloud backends without lock-in.

Core outcomes

Vaner is successful when it improves one or more of:

• answer quality
• prompt-time latency
• cost per good answer

Start here

• New users: Getting started
• Concepts and terminology: Concepts
• Setup and tuning: Configuration
• Integration modes: Integrations