Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Python agentaianthropicclaude-codecompressioncontext-engineeringcontext-windowcursorfastapilangchainllmmcpopenaiprompt-engineeringproxypythonragtoken-optimizationtokenstypescript
feat: export CompressionCache from cache package
chopratejasc125939a9e3e
diff --git a/headroom/cache/__init__.py b/headroom/cache/__init__.pyindex 65fa912fc..d92c16b47 100644--- a/headroom/cache/__init__.py+++ b/headroom/cache/__init__.py@@ -36,6 +36,7 @@ from .base import (CacheStrategy,OptimizationContext,)+from .compression_cache import CompressionCachefrom .dynamic_detector import (DetectorConfig,DynamicCategory,@@ -74,6 +75,8 @@ __all__ = [# Semantic caching"SemanticCacheLayer","SemanticCache",+ # Compression cache (token headroom mode)+ "CompressionCache",# Prefix cache tracking"PrefixCacheTracker","PrefixFreezeConfig",