Film recorded with an earlier version; the written reference is current.
Where it is
The packaged OmaRAW skill · omaraw skill
Quick overview
On Omarchy, installing the package links its included skill for the identified user. Start a new agent session after installation.
Run omaraw skill to inspect the links. Run omaraw skill ‑‑link to repair them or set them up for another user. Custom skills are preserved.
Close every OmaRAW window before an agent opens a catalog through the command line. Name the catalog and photographs you mean.
Ask for a concrete result, such as warming selected portraits slightly or exporting your three-star picks at 2048 pixels.
The agent should inspect the existing edit, use the app’s operations, export a preview and review the result. Stars, flags and colour labels remain your choices in Library.
The package includes the OmaRAW skillStart a new agent session after installation. Close OmaRAW before CLI work.
Inspect before making changesThese commands ran with the OmaRAW window closed in a separate demo catalog.
Apply a small exposure adjustmentThe installed skill guides the agent to use the app’s actual operations.
Read the edit backA successful command should be checked against the resulting settings.
Export the sample collectionActual result: six JPEGs, at most 2048 pixels on the long edge.
Review the photograph as well as the numbersExample request: brighten this photo slightly, then export web JPEGs.
What the packaged skill enables
The OmaRAW package includes instructions for an agent to use the application's headless commands. An agent can inspect an edit, make supported adjustments, organise photos and export results through the same processing engine. Installing the skill does not give it an automatic understanding of which photographs you prefer.
On Omarchy, installation links the skill for the identified user. Start a new agent session so it can discover the skill. omaraw skill checks the links; omaraw skill --link repairs or creates them for the current user while preserving custom skills. The included command reference below gives concrete syntax.
Make a useful request
Name the catalog and the intended photos or collection, then describe a visible result and any output requirements. For example: “Use my Portraits album, warm the selected photos slightly, keep the skin natural, and export sRGB JPEGs at 2048 pixels.” A clear destination and naming request avoids ambiguity about where finished files belong.
Close every OmaRAW window before the agent opens that catalog through the CLI. The agent should inspect existing settings, apply supported operations, read the result back and export a preview for visual review. A successful JSON response confirms a command, not that a photograph looks good.
Decisions you still make
Choose ratings, flags and colour labels in Library. An agent can use those choices to find a set, but this workflow does not ask it to invent your culling decisions. Review masks, crops, faces and strongly changed colours before accepting a large batch. Keep the source catalog and original photos backed up.
What the film verifies
The recorded example uses real headless commands in a separate demonstration catalog: inspect a photo, apply +0.25 EV, inspect the saved value, and export six JPEGs at no more than 2048 pixels on the long edge. It then opens the edited photograph for review. The terminal output comes from those actual commands; it is not an invented agent conversation.
Check the installed skill
omaraw skill
omaraw skill --link
The first command reports its location and links. The second links the packaged skill for your user, preserving custom copies. Start a new agent session afterwards. If a package is installed without an identifiable user, run the linking command from your own account.
Give the agent a clear task
Close OmaRAW first, then name the catalog and selection. For example: “In my Travel catalog, export my three-star picks as sRGB JPEGs, 2048 pixels on the long edge, into the Web exports folder.”
The agent can read your ratings and flags. It does not make culling decisions for you.
Command-line reference
The package installs the full reference at /usr/share/doc/omaraw/cli.md. The source copy is on GitHub, accessible while the repository is private only to people with access.