Offline-first · Multi-lens · 508-ready

Discovery without guesswork.

A single-file HTML5 workbench that ingests documents locally, applies the Apex Lotus lenses, builds a grounded prompt pack, and exports a clean Markdown report.

Local document intake

PDF, DOCX, MD, TXT, JSON, CSV, code, and image metadata stay in browser memory until you choose a model endpoint.

Four strict lenses

Learning Architect, JurisLogic, Forensic Linguist, and QA / Secure Code outputs use fixed structures.

Visible execution loop

Triage, lens selection, parsing, audit, and synthesis are shown as a traceable pipeline.

Accessible by design

Skip link, keyboard controls, large touch targets, reduced-motion support, and pure black/white contrast mode.

Discovery Console

Ingest → Ground → Analyze → Export

Drop context files here

PDF and DOCX parsing runs client-side. Text and code load through FileReader. Images are listed as evidence objects.

Embedded Source Pack

Loaded into this single HTML file

1TriageIdentify material type and analysis need.
2Lens SelectionApply the chosen professional frame.
3ParseExtract assertions, claims, code logic, or evidence units.
4AuditCompare extracted content to source rules.
5SynthesizeProduce the final Markdown report.

Analysis Output

Waiting for payload

Awaiting analysis.

Load context, choose a lens, then run the offline pass.

Prompt Pack

Embedded source docs + selected lens + payload

ReadyWorkbench loaded.