Siri & Apple Intelligence
On a Mac, Social Archiver plugs into Apple's own stack twice over. Apple Intelligence can write AI comments, suggest tags, and pull out places without the post ever leaving the machine. Siri, Spotlight, and Shortcuts can reach into your library by name: search it, open a tag, open an author, jump to a section, or hand it a link to save.
Apple platforms only
Everything here is macOS, plus the iPhone and iPad parity noted at the end. Windows and Linux desktops keep the AI CLI and Cloud AI paths described in AI Comments.
What you need
| Feature | Requirements |
|---|---|
| Shortcuts actions — Save Link, Open Section, Open Tag, Open Author | macOS 13 or later, desktop app 0.7.0+ (tag, author, and section actions: 0.7.2+) |
| Your archives in Spotlight | macOS 15 or later, desktop app 0.7.0+ |
| Opening an archive from a Spotlight result | macOS 27 |
| Spoken Siri requests | macOS 27 with Apple Intelligence turned on |
| Apple Intelligence for AI comments, tag suggestions, and place extraction | Apple silicon, macOS 26 or later, Apple Intelligence turned on. Desktop app 0.6.16+, or the social-archiver CLI 0.1.15+ |
There is nothing to switch on for the Siri half: install the desktop app, launch it once, and the actions and the Spotlight index appear on their own.
Ask Siri
With the desktop app installed, Siri can work on your library out loud:
- "Search Social Archiver Desktop for kimchi stew" — opens the app with the library in search mode for that term.
- "Open the kimchi stew recipe I saved" — Siri resolves the archive itself from the Spotlight index and opens it.
Spoken "open …" phrases usually become a search
Once an app offers in-app search, Apple Intelligence claims most "… in app name" sentences for it — "Open Reels in Social Archiver Desktop" is heard as a search for "Reels". The Open Section, Open Tag, and Open Author actions still run exactly as written from the Shortcuts app, Spotlight, or a keyboard shortcut you assign to them.
Siri also sometimes answers from its own domains (Mail's archive search, Find My) when a request sounds like one of theirs. Naming the app — "in Social Archiver Desktop" — is what steers it back.
Spotlight
The desktop app donates your archives to Spotlight in the background. Each item carries its title, author, platform, summary, and body text, so a phrase from the middle of a post finds it. Selecting a result opens that archive in the app.
- The first index runs shortly after the app launches and covers the whole library at once — about a second for a thousand archives. After that it keeps up as you archive, tag, and delete.
- Indexing is automatic — there is no setting for it.
- Seeing nothing? Launch the desktop app at least once, then check that Social Archiver Desktop is enabled under System Settings → Spotlight → Results from Apps.
Shortcuts actions
Open the Shortcuts app and search for Social Archiver Desktop to find these actions:
| Action | Takes | What it does |
|---|---|---|
| Save Link | A URL | Opens Quick Capture with the link filled in. Nothing is saved until you confirm it. |
| Open Section | Library, Reels, Tags, Places, Shopping, Subscriptions, Review, or Settings | Switches the app to that section |
| Open Tag | One of your tags | Opens the library filtered to that tag |
| Open Author | One of your authors | Opens that author's page |
The tag and author pickers suggest your recent tags and most-archived authors, and typing matches any of the others by name. Each row shows how many archives it holds, so duplicates are easy to tell apart.
Build a shortcut once and you can run it from the Shortcuts menu bar item, from Spotlight, or with a keyboard shortcut you assign in Shortcuts — no dictation involved.
A shortcut that fails right after an app update
The desktop app updates itself in the background, but macOS may keep the previous copy of its Shortcuts helper running, which then reports that the action no longer exists. Quit Social Archiver Desktop and open it again.
Apple Intelligence for AI comments
Apple's built-in Foundation Model runs a comment entirely on the Mac: no internet, no account, no credits, and the post never leaves the machine. On a Mac it covers Summary, Key points, Glossary, Reformat, Translation, Custom, tag suggestions, and place extraction; Fact check always needs web search, so it stays on Desktop or Cloud AI. AI Comments has the full matrix and how the target picker chooses.
In the desktop app
Settings → Local executor → AI comment executor lists an Apple Intelligence row next to your AI CLIs, described as On-device model · no account, no credits. Its status tells you exactly what is missing when it is not usable: Ready, Preparing model…, Turn on Apple Intelligence, Requires macOS 26, or Requires Apple silicon.
In Obsidian on a Mac
The plugin's built-in executor runs AI CLIs only — it has no on-device model of its own. Turn on Settings → Social Archiver → AI Comments → Run through the Social Archiver CLI (plugin 4.7.8 or newer) and the standalone CLI serves that vault instead, with Apple Intelligence alongside Claude, Gemini, and Codex.
It needs social-archiver 0.1.15 or newer on your PATH:
brew install hyungyunlim/tap/social-archiver-cliThe desktop app can install the same tool from Settings → Command line → Install ‘social-archiver’ command. Once the CLI is found, the plugin setting shows its version, whether its executor is running for this vault, and the providers it can see — Apple Intelligence included. See Command Line (CLI) for the rest of the tool.
Checking it from a terminal
social-archiver executor --providers --format textApple Intelligence appears as apple, with the context window it reports and, when it is not ready, the reason.
The first run says "Preparing model…"
macOS does not download the on-device model until something asks for it, and a fresh install or a major macOS upgrade starts over. The quickest way to trigger it is to open Writing Tools once anywhere on the Mac — select some text, right-click, and choose a summary — then leave the Mac on power and Wi-Fi. The download takes roughly half an hour, and Apple Intelligence reports itself unavailable until it finishes.
Length limit
The on-device model shares about 4,096 tokens between your prompt and its answer (Korean runs near four characters per token). A post that does not fit is refused rather than quietly truncated — run those on Desktop or Cloud AI instead. The first request after a restart also takes a few seconds longer while the system loads the model.
Also on-device: text in images
Settings → Image Text → Recognize text in images reads text out of archived images with Apple Vision on the same Mac, and never uploads them. With Include image text automatically on, that text is recognized before AI comment and place-extraction requests, so a screenshot's contents reach the model along with the post. This is Apple's on-device vision stack rather than Apple Intelligence, so it works on any Mac the app supports.
On iPhone and iPad
The mobile app has the same two halves:
- Siri and Spotlight — since Social Archiver 2.2.11 on iOS 27, archives appear in Spotlight and open from a result, Siri can search your library or open an archive you describe, and a Save Link action is available in Shortcuts.
- Apple Intelligence — on iOS 26 or later it runs Summary, Glossary, Custom, and tag suggestions on the device. See AI Comments for what stays on Desktop or Cloud AI, and Mobile App for the rest of the app.