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
Commits on main
fix: parse actual token usage from SSE streams instead of byte estimation
8c22a72 Tejas Chopra committed 9 months ago
fix: account for cached tokens in cost calculation across all providers
6bb35eb Tejas Chopra committed 9 months ago
Add E2E tests for Google multimodal content preservation
62ee1dc chopratejas committed 10 months ago
Fix Google Gemini conversion to preserve non-text content (Gap #4)
dd4c65e chopratejas committed 10 months ago
Update compressor CCR markers and add batch compression eval
cec2b76 chopratejas committed 10 months ago
Add multi-provider batch API support with CCR post-processing
95fd6d8 chopratejas committed 10 months ago
Update memory documentation for HierarchicalMemory system
8c88d49 chopratejas committed 10 months ago
Add comprehensive evaluation framework for compression accuracy
92e4a24 chopratejas committed 10 months ago
Update test for lowered TOIN confidence threshold default
fa5ba8f chopratejas committed 10 months ago
Add reasoning capability forwarding to HeadroomAgnoModel
4ac48ed chopratejas committed 10 months ago
Add pluggable storage backend abstraction for CompressionStore
3ffe686 chopratejas committed 10 months ago
Integrate DynamicContentDetector into CacheAligner (Phase 1)
2ce2643 chopratejas committed 10 months ago
Fix cost tracking to use LiteLLM pricing database correctly
6ec85f3 Tejas Chopra committed 10 months ago
Add quality retention eval and fix linting for Python 3.12
bd2d447 chopratejas committed 10 months ago