Slimcontext Mcp Server

Wraps the SlimContext library to compress chat histories when they approach token limits. You get two tools: trim_messages removes the oldest messages while preserving system prompts and recent context, while summarize_messages uses OpenAI's API to condense the middle of conversations into concise summaries. Both let you configure token thresholds, how many recent messages to preserve, and when compression kicks in. The trimming strategy is fast and deterministic, good for quick cleanup. The summarization route costs API calls but maintains semantic context better for long conversations. Useful when you're building chatbots or agents that need to stay under model context windows without losing conversational thread.