14-stage Fusion Pipeline for LLM token compression — reversible compression, AST-aware code analysis, intelligent content routing. Zero LLM inference cost. MIT licensed.
-
Updated
Apr 1, 2026 - Python
14-stage Fusion Pipeline for LLM token compression — reversible compression, AST-aware code analysis, intelligent content routing. Zero LLM inference cost. MIT licensed.
Windowed, reversible context management for DeepSeek Harness (DSH): compression, historical retrieval, and native compaction integration.
Automate content research, card news, images, voice, and video from one prompt with an end-to-end Claude Code content pipeline
Context compression for agent CLIs: evidence is never cut, every changed output stays recoverable, and the measurement refuses rather than guesses.
To associate your repository with the reversible-compression topic, visit your repo's landing page and select "manage topics."