Ctakes_USD wraps the curated cTAKES 6 bundle. The normal site run is one command: 
bash scripts/run_pipeline.sh -i input/batch1 -o outputs/batch1
That command is doing more than it looks like. It finds Java, finds the bundled cTAKES runtime, picks the STAR pipeline, rewrites the dictionary XML so cTAKES can find the local HSQL dictionary files, builds the helper Java classes, strips the heavy writers out of the .piper file, adds the SQLite writer, runs cTAKES, and writes the run receipt. Each page below covers one of those pieces.

Setup and files

What setup.sh downloads, the folders a run touches, and how input notes must be named.

Run recipes

The exact command for each kind of run, from the normal database run to HTML debug views.

Pipelines and writers

The pipeline keys, the stages inside the default STAR pipeline, and what each writer profile keeps.

Dictionary

The dictionary XML the run uses, how the wrapper rewrites it, and what belongs in dictionaries/.

Outputs

The annotations.sqlite tables, the CSV/XMI folders, and the run manifest fields.

Operations

Autoscale knobs, offline kits, the 100-note validation pack, and quick fixes for common failures.
If the person running the batch only needs commands, send them the run recipe instead.