DOCUMENTATION / VERSION 0.3.16

START LOCAL. FIND EVERYTHING.

Setup, indexing, native search and troubleshooting — documented clearly.

SEARCH DOCUMENTATION

DOCUMENTATION INDEX

01

QUICK START

Get NOMEN ready, add the first folder and run an incremental local index.

02

ADDING SOURCES

Add existing folders, check several sources and understand how NOMEN reads them.

03

CLIP + VLM

Use fast local CLIP tags, then optionally add richer descriptions with Ollama.

04

SPOTLIGHT / NATIVE SEARCH

NOMEN writes kMDItemKeywords; macOS ingests them asynchronously.

05

CATALOG + METADATA

See exactly what is stored in <source>/__catalog__ and on each file.

06

TROUBLESHOOTING

Separate catalog, metadata-write and system-ingestion issues before repairing anything.

No matching topic. Try “VLM”, “source”, “Spotlight” or “catalog”.

QUICK START: FOUR STEPS.

The normal workflow is short. Heavy processing is incremental after the first index.

01ADD A SOURCE

Choose a folder containing the images and documents you want searchable.

02CHECK THE SOURCE

New sources are checked automatically. Check several folders for a sequential job.

03PRESS INDEX

NOMEN scans, processes and writes metadata. Optional VLM work may take longer.

04SEARCH IN FINDER

After macOS ingests the metadata, search naturally for contents absent from filenames.

CLIP + OPTIONAL VLM

FAST BASELINE. PRECISE UPGRADE.

A successful VLM description becomes the primary source of image keywords. CLIP remains stored locally and acts as fallback.

  • CLIP IS INCREMENTAL
  • VLM RESULTS ARE CACHED
  • WEBP IS TRANSCODED IN MEMORY
  • 20–50 KEYWORD MAXIMUM
  • STOP UNLOADS OLLAMA SAFELY

SUPPORT + FAQ.

Start with the system layer that is actually failing. Avoid full Spotlight resets during routine indexing.

DIRECT SUPPORT

EMAIL SUPPORT REPORT A PROBLEM

VERSION 0.3.16 / macOS PRIVATE BETA

Does NOMEN move my files?

No. It reads files where they already live, stores derived catalog data beside the source and writes native searchable metadata.

Do I need Ollama?

Only for optional richer VLM descriptions. CLIP image indexing and document extraction work without Ollama.

Can I stop and resume indexing?

Yes. STOP completes the current safe unit of work, unloads the VLM and publishes results already completed. Later indexing reuses cached work.

Where are descriptions stored?

Successful image descriptions are cached as local JSON under <source>/__catalog__/vlm_desc/.

What happens when I remove a source?

NOMEN removes the folder from its registered source list. Source files are not deleted. Destructive catalog cleanup is a separate, explicit operation.

When will Windows be available?

The Windows version is planned after equivalent native-metadata injection, packaging and release tests are complete.

NEED A REAL ANSWER?

CONTACT SUPPORT