context-optimizer

What problem does it solve?

The Context Optimizer solves context window overflow for AI agents operating within a 64k token limit by automatically pruning, compressing, and archiving conversation history while preserving high-priority and recent content.

Core Features & Use Cases

• Adaptive pruning: combines semantic, temporal, and extractive techniques to reduce token usage without sacrificing essential context.
• Dynamic context: scores content relevance to current goals or queries and adjusts the active memory in real time.
• Hierarchical memory: stores pruned material in an archive for on-demand retrieval, enabling long-running conversations.
• Real-time monitoring: tracks health, token usage, and archive statistics to maintain stable performance.

Quick Start

1. Install dependencies: npm install
2. Initialize the pruner in your app and call createContextPruner with 64k context limits.
3. Process messages with processMessages and optionally retrieve related content from the archive for enhanced context.