Corpus & Engine Statistics

Total Archive Docs
35,455
1.72 GB (37.7 GB raw)
Embedded Documents
30,264
1.23 GB (Optimized APFS)
High-Signal Chunks
199,998
1.23 GB (884k pruned)
Avg Chunks / Doc
6.6
High-signal passages
Corpus Date Span
1970 – 2026
Peak: 2000 – 2023 (82%)
Corpus Storage & Index Footprint
Source Documents (Original Archive): 37.66 GB
Search Catalog (Converted Markdown): 1.72 GB
Compact Vector Store (ChromaDB): 1.23 GB (-12.95 GB saved)
Okapi BM25 Lexical Index (35k Docs): 321.06 MB
Master Entity & Keyword Index: 60.68 MB
Specialized Domain Dictionaries: 4 Indexes (156.4 MB)
Local Apple Metal LLM Weights (GGUF): 1.88 GB
Why Inspect the RAG Pipeline Directly?

When monitoring and auditing offline AI search across state records, checking the Corpus Diagnostics provides several key operational guarantees:

  • Storage Efficiency: Confirms the 91.3% compaction ratio from 14.18 GB down to 1.23 GB on local APFS storage.
  • Zero Hallucination Verification: Inspect raw document passage coverage before LLM synthesis.
  • Corpus Integrity: Verifies 35,455 historical and active documents span cleanly across 1970–2026.
Open Side-by-Side Algorithm Benchmark Export Full Metadata JSON