| Page | Description |
|---|---|
| API Reference | All exports, types, options, and result fields |
| Compression Pipeline | How compression works: classify, dedup, merge, summarize, guard |
| Deduplication | Exact + fuzzy dedup algorithms, tuning thresholds |
| Token Budget | Budget-driven compression, binary search, custom tokenizers |
| LLM Integration | Provider examples: Claude, OpenAI, Gemini, Grok, Ollama |
| Round-trip | Lossless compress/uncompress, VerbatimMap, atomicity |
| Provenance | _cce_original metadata, summary_id, parent_ids |
| Preservation Rules | What gets preserved, classification tiers, code-aware splitting |
| Benchmarks | Running benchmarks, LLM comparison, interpreting results |