NIMOS / ENGINES / Documentation

← THE BUILD SYSTEM

Documentation

THE MEMORY

Document & Context Ingestion

The layer that turns PDFs, Word files, spreadsheets and web pages into clean, searchable context.

DRAWN BY THE HOUSE

One mark per engine, on one grid, in one hand.

5

TOOLS

MIXED

WIRE

N/A

THIS BUILD

No document-ingestion service runs in this site.

THE PROOF LEDGER →

WHAT

What it is

Five entries: LlamaIndex core (JS and Python), MarkItDown, pdf-parse, mammoth and the v0 SDK lane.

WHY

Why it exists

Client assets arrive in every format. A build should read them as structured evidence, not fight a file extension for an afternoon.

HOW

How it runs

The shared pyenv and the MarkItDown CLI; npm parsers in projects; corpora chunked with LlamaIndex and registered with source, chunk count and freshness.

WHEN

When it runs

Documents arrive, a corpus is built, a RAG lane is needed, or a spec must be ingested before a build.

WHO

Who owns it

CEO owns the archive; Research watches parsing; Build writes ingest flows; Data keeps the corpus register; Ops runs the ingest queue.

TOOLS

The registry

The engine manifest is the source of truth. These are its entries, including the ones that are not used on this build — the proof ledger says which is which.

@llamaindex/core

npm

Deep context parsing — split complex code markdown and documents into clean chunks.

@microsoft/markitdown

pypi

Converts incoming client assets (PDF, Word, Excel) into clean markdown agents parse instantly.

pdf-parse

npm

Zero-dependency text arrays straight out of PDF layouts.

mammoth

npm

Reliable .docx→HTML conversion — pure structured content, styling junk stripped.

v0.dev API

apisource ↗

Stream structured layouts from Vercel's generative API into workspace context.

5 registry entries. The proof ledger on the Gantry page says which of them this build actually used.

RUN

Run it

ENGCTL / THE ENGINE FLOOR

$ engctl show documentation$ engctl awaken documentation$ engctl install documentation --dir ./site

`engctl` is the registry control surface. `awaken` checks wiring before it activates a cell; `install` writes the parts into a project.

PROOF

On this build

No document-ingestion service runs in this site.

N/A and PARTIAL are stated, not hidden. The ledger is generated from this project, not from a template.

LIMITS

What this does not claim

N/A here. This site does not run an ingestion service; the library lives in the studio.

The engine is the how. The brief is the start.

Documentation — the the memory | Nimos