Simulation run on this device. Scenario values are not sent anywhere — the engine is fully local.
agents studio · pro
PDF-to-RAG Readiness Scan
Score a PDF's layout for extraction risk before it poisons the index.
Is this document chunkable — or a table soup?
Engine 1.0.0 · studio
- Regions detected
- 5
- Extraction risk
- 7/10
- Recommended chunking
- layout-aware + table serialization
- highMulti-column text
use a layout-aware parser; naive extraction interleaves columns
- highWide tables
detect + serialize tables to markdown; never chunk mid-table
- mediumFigures & captions
keep caption attached to its figure chunk; OCR only if text is in the image
- mediumFootnotes
link footnotes to their reference, do not emit as standalone chunks
- highEquations
extract LaTeX where possible; a lost symbol flips the meaning
Parse for the riskiest region first — it determines whether the rest of the pipeline is trustworthy.
Method
- Scores extraction risk per described layout region (columns, tables, figures, footnotes, equations, running headers) and recommends a chunking strategy.
- Risky regions are parsed first — they decide whether the rest of the pipeline can be trusted.