spinosa

GitHub
Home / Docs / Glossary

Glossary

Plain-English definitions of terms you'll encounter in Spinosa.

Plain-English definitions of the terms you see most often. This page stays short on purpose. For full explanations, follow the linked canonical pages.

Core terms

Term Meaning
Corpus Your source collection: PDFs, transcripts, notes, images, and related research files. See Corpus Structure.
Workspace The local folder Spinosa creates from your corpus, including raw/, maps/, system/, and agent_reports/. See Corpus Structure.
Orchestrator The top-level coordinator that decides which agents should run for a request. See Agents & Pipeline.
Pipeline The sequence of agent steps used to answer a given request. See Agents & Pipeline.
Sub-agent A specialized helper such as Searcher, Analyst, Writer, or Verifier. See Agents & Pipeline.

Workspace terms

Term Meaning
raw/ Converted local copies of your source documents. This is the evidence layer agents search and cite.
maps/ Navigation maps that help agents find clusters of relevant files and themes.
system/ Settings and index files such as configuration.md, context.md, dictionary.md, and workspace_index.md.
Dictionary The vocabulary list Spinosa builds from the corpus: names, places, organizations, concepts, aliases, and uncertain terms.
YAML header The metadata block at the top of a raw/ file describing source type, language, dates, people, topics, and related fields.
Source intake The workflow for adding new material to an existing workspace without rebuilding everything from scratch.

Report terms

Term Meaning
agent_reports/ The folder where startup reports, answers, and other report artifacts are written.
Evidence packet An intermediate file that packages candidate quotes and source paths for report drafting.
Confidence level A signal of how directly a passage supports a point: typically high, medium, or low.
Verification status The report outcome after source checking: pending, verified, corrected, or failed. See Reports & Charts.

Conversion terms

Term Meaning
OCR Optical Character Recognition: turning scanned or image-based text into searchable text.
MarkItDown The converter used for many office, EPUB, HTML, and text-based PDF formats.
RapidOCR The local OCR engine used for scanned PDFs and image-heavy files.
Obsidian wikilinks Double-bracket links like [[filename]] used in navigation maps for graph-style browsing in Obsidian.

Next reads