links.md lists two autogenerated preserved formats; a fresh account preserves four plus a preview

[truecourse-agent]

The "Each Link consists of the following data" table lists exactly two autogenerated preserved formats. A fresh account preserves four, plus a preview that is always generated. The other surfaces that state a count disagree with this page and with each other.

The documented promise

docs/usage/links.md:60-61 — the only two Autogenerated rows in the table:

| Screenshot | Autogenerated | Will be autogenerated. |
| PDF | Autogenerated | Will be autogenerated. |

There is no row for Webpage (single-file HTML), Readable, the preview image, or the Archive.org snapshot.

What a fresh account actually does

Checked against linkwarden/linkwarden v2.16.1 (7550ebb8907e0cd22e8b3e0785bf3da03ab96e3a), on a brand-new account with nothing changed.

GET /api/v1/users/me:

archiveAsScreenshot: true    archiveAsMonolith: true
archiveAsPDF:        true    archiveAsReadable: true
archiveAsWaybackMachine: false

Four of the five archive settings are on by default. Settings → Preferences shows the same five checkboxes with the same four ticked:

And one link preserved end to end on a stock instance carries all four plus the preview:

"image":    "archives/1/2.jpeg"
"pdf":      "archives/1/2.pdf"
"monolith": "archives/1/2.html"
"readable": "archives/1/2_readability.json"
"preview":  "archives/preview/1/2.jpeg"

Reading a fresh account's first preserved link back and asserting the table's two-format promise fails on both extra formats:

• expected monolith to be null; actual "archives/1/1.html"
• expected readable to be null; actual "archives/1/1_readability.json"

The surfaces also disagree with each other

• docs/usage/links.md:60-61 — two (Screenshot, PDF)
• linkwarden/linkwarden README.md:32 — "saves a copy of each webpage as a Screenshot and PDF" — two
• linkwarden/linkwarden README.md:43 — "Auto capture a screenshot, PDF, and single html file of each webpage" — three
• docs/getting-started/welcome.md:10 — "saves a copy of each webpage in multiple formats" — unspecified

So the count a reader comes away with depends on which page they land on, and none of them matches the four-plus-a-preview a stock install produces.

Suggested fix

Give the table a row per format that is generated by default — Screenshot, Webpage (single HTML), PDF, Readable, and the preview image — and note that the Archive.org snapshot is the one that is off by default and opt-in. README.md:32 in the main repo would need the same correction to agree with README.md:43.