フォルダの自動振り分け: what the free tier covers

Searching for what フォルダの自動振り分け costs turns up two contradictory answers. One says it is free, because macOS already contains the machinery. The other says a paid tool is required. Both are accurate, and the reason they coexist is that the free route charges in a currency that does not appear on an invoice. What follows separates the two: where the free ceiling actually sits, what the paid tools charge as of 2026, and which of the two costs is larger for a given folder.

Every price below was read from the vendor's own store page rather than from a review.

The ceiling on free is higher than most people assume

Everything in this table ships with macOS. There is nothing to buy, register, or trial.

Included mechanism	Cost	Moves files
Desktop stacks	Free	No
Saved searches (smart folders)	Free	No
Finder tags	Free	No
Folder Actions	Free	Yes
Automator	Free	Yes
Shortcuts, including its command line tool	Free	Yes
Launch agents, watching paths or running on an interval	Free	Yes
Standard shell tools and the Spotlight command line tools	Free	Yes

Five of those move files. The premise that automating a Downloads folder starts with a purchase is simply wrong, and it is worth establishing that before looking at any price.

Folder Actions in particular is further along than its reputation suggests. Thirteen ready made scripts are installed at /Library/Scripts/Folder Action Scripts/, covering image rotation, format conversion, and alerting on new arrivals, and the configuration app is at /System/Library/CoreServices/Applications/Folder Actions Setup.app. Attaching a sample to a test folder takes a few minutes and proves the whole chain works before any custom logic is written.

Shortcuts adds a visual editor on top of the same capability, and the shortcuts command installed with it accepts run, list, view and sign. That makes it possible to build the decision logic in a graphical editor and let a launch agent decide when it fires.

The line between free and paid is not a feature line. It is whether the conditions can be written out in text.

Three specific things are missing from the included set, and they are worth naming because they are what the money buys. The first is a list view of the rules. Rules built from the included pieces end up distributed across scripts, property lists and shortcut definitions, so answering the question "what is currently running against this folder" means going and looking. The second is a preview. None of the included mechanisms offer a way to evaluate a rule and see the outcome without performing it, so that safety net has to be built by hand. The third is a log of what happened. Knowing which rule touched which file after the fact requires writing that record explicitly.

At five rules none of the three matter, because the whole setup fits in one person's head. Past a dozen they matter a great deal, and the point where a self built system stops being maintained is usually the point where nobody is certain any more what it is doing.

What the paid tools charge in 2026

Two applications dominate this category on the Mac, and one open source option sits alongside them. All prices are one time purchases rather than subscriptions.

Tool	Price	What it covers
Hazel 6	42 US dollars	One license
Hazel 6 Family Pack	65 US dollars	Up to five people in one household
Hazel 6 upgrade	20 US dollars	Moving from a previous version
Keyboard Maestro 11	36 US dollars	One license
Keyboard Maestro 11 upgrade	25 US dollars	Moving from a previous version
organize	Free	Open source command line tool

Hazel is at version 6 and is published by Noodlesoft, LLC. The store page carries the family pack restriction in plain language, limiting it to five members of a single private household, and a free trial is offered alongside the purchase links.

Keyboard Maestro states its licensing model directly on its main page.

Keyboard Maestro 11 is a one-time purchase. No subscription, you can purchase it now for US$36 plus applicable tax and use version 11 indefinitely. Source: keyboardmaestro.com

The third option is free and positions itself explicitly against the paid ones. Its documentation describes organize as a command line, open source alternative to apps like Hazel. Rules are written in YAML, and organize sim runs the whole rule set without touching a single file, printing what would have happened. That dry run capability is the single most useful safety feature in this entire category, paid or free.

Lining the three up, the price tracks the input method rather than the capability. Tools with a visual rule editor carry a price. Tools where rules are typed into a configuration file do not.

The listed price does not include the next major version

Both paid tools list an upgrade price separately, which is the detail that decides total cost over several years. Hazel charges 20 US dollars to move from an earlier version, Keyboard Maestro charges 25.

How often a major version appears is the vendor's decision, so the multi year total cannot be fixed in advance. What can be stated is the shape of it: nothing is charged automatically, and continuing to run the version already purchased costs nothing further. That is a meaningfully different risk profile from a subscription, and it is the reason both vendors say so prominently.

There is a second thing the price does not cover, which is keeping up with macOS. Sorting rules depend on file access permissions, and permission handling changes between major releases. A paid vendor absorbs that maintenance, but the number of future macOS releases covered by today's purchase is not stated anywhere, because it cannot be.

Putting a multi year figure on it requires one assumption, which is how many major versions get bought. Assuming one upgrade during the period, Hazel totals 42 plus 20, or 62 US dollars. Keyboard Maestro totals 36 plus 25, or 61. Skipping the upgrade and continuing on the purchased version is a legitimate option that costs nothing, so those figures are a ceiling rather than a forecast. For a household running several Macs, the Hazel family pack at 65 dollars covering up to five people changes the per machine arithmetic considerably.

Both vendors also offer a trial before purchase, and the Noodlesoft store page links its refund policy alongside the buy buttons. That matters more than it sounds, because whether a rule engine is worth 42 dollars depends entirely on what is sitting in the folder. Running a trial against the actual mess for a few days produces a clearer answer than any feature comparison.

What the free route charges instead

Choosing the included tools has a cost. It is just denominated in hours, and it arrives in three separate installments.

The first is writing the conditions. Every decision has to be spelled out: which extensions go where, whether the relevant date is created or modified or added, and what happens when a file with that name already exists at the destination. That last one is where most self built rule sets lose data, because the obvious implementation overwrites.

The second is repair. After a macOS update stops something working, the cause has to be isolated by hand. It could be a permission that was revoked, a path that changed, or an event that was missed. A visual tool narrows this because the rules are visible as a list; a shell script does not.

The third is recall. Six months later, adding a rule starts with rereading what was written. A configuration file with twelve rules in it is a small piece of software, and it has the same maintenance characteristics as any other small piece of software written once and never touched again.

Paid tools absorb the first and third of those. Neither side eliminates the second.

The conditions under which free is clearly the right call are specific: the rule set stays at roughly five rules, it is not expected to grow, and the person maintaining it already works comfortably in a terminal. Outside those conditions, the arithmetic tips.

One practical way to avoid deciding prematurely is to defer the purchase by a month. Write three rules using the included mechanisms, run them against the live folder, and keep a note of every rule that felt worth adding but was not written. After four weeks the note answers the question directly. If it is empty, a paid tool would sit unused. If it has ten entries on it, the rules are already past the point where a text based configuration stays readable, and the price of a tool that lists them on one screen is easy to justify. The month is not wasted either way, because the three rules keep working regardless of what gets decided afterwards.

Two smaller points do not appear on any price page. These tools are priced in US dollars, so the amount actually charged outside the United States depends on the exchange rate at the moment of payment and on how the card issuer handles the conversion. And a purchase covers the application, not the rules; the time spent designing the rule set transfers to whichever tool runs it, which means that work is never wasted even if the tool changes later.

Two numbers that decide it before any price is compared

Comparing feature lists is slower and less conclusive than counting two things in the folder that is actually causing the problem.

The first is the extension spread. On one Mac with a genuinely overgrown Downloads folder, the distribution ran 197 PNG files, 165 PDFs, 137 CSVs, 84 MP4s and 66 DOCX files, with images further split across 57 jpeg, 42 jpg and 37 heic. A concentrated distribution means a handful of rules cover most of the volume, which favors the free route. A long tail means many rules, which favors a tool where rules are easy to read back.

The second is provenance coverage. On the same machine, 383 of 1,062 files carried the extended attribute recording where the file came from, which is 36 percent. Screenshots, AirDrop transfers, copies from external disks and files written directly by applications carry nothing. Anyone considering a paid tool specifically for its ability to sort by originating website should measure this first, because that headline capability will be silent on the majority of files in a typical folder.

Neither number takes more than a few minutes to obtain, and together they answer the purchasing question more decisively than any comparison table. The pricing page sets out how this kind of tool is priced when the terminal is part of it, and the comparison places that next to the rule engine and dual pane categories.

What to change first

Count the extension spread and the provenance share in the one folder that is actually causing the problem, then decide. A concentrated folder with a handful of file types needs no purchase at all. If the part that keeps stalling is the loop of editing a rule, checking the folder window, and confirming the result in a terminal, that specific gap is what Atriens was built to close.