My import came in with warnings
None of them stopped the import. Each means a node landed slightly poorer than intended — and one of them is worth going back for.
The rule behind all of them
Every unrecognised value in a paste degrades safely. An undeclared group lands ungrouped. An undeclared relation falls back to a plain edge. A malformed date is ignored. In each case the node itself is fine and it is in the right place, so refusing the whole import would cost you more than the missing attribute does.
There is one exception, and it is the exception that proves the rule.
lens: naming a lens you do not have is an error, not a warning. Every other bad value costs the node an attribute; a bad lens costs it its whole location — it would land correctly formed in whichever lens you happened to have open, with nothing on screen to say so. Silent misplacement is worse than a refusal you can act on, so the importer refuses and suggests the nearest lens you do own.Go back for this one
Link target "…" doesn't resolve
The edge is dropped. Nothing after the import will tell you — a dropped edge and an edge nobody wrote look identical — so this warning is the only place it is visible.
If the target looks like a longer, more spelled-out version of a node you already have (wg-richard-nixon where you have wg-nixon), the model reconstructed a slug instead of copying it, and that is worth fixing before you paste. My links didn't connect anything is the full diagnosis.
If the target is a node a later batch will create, this is expected and correct — a forward reference fills itself in when that node arrives.
The rest, in one table
| Warning | What happened |
|---|---|
| Group “…” isn't declared in this lens | The node imported ungrouped and took no colour. Either add that group in lens settings and re-assign, or fix the value and paste again. Nothing is lost either way — grouping is a property you can change afterwards. |
| No group set | Same outcome, no opinion offered. Ungrouped is a legitimate place for a node to be. |
| Kind “…” isn't a declared facet value | Only appears in lenses that constrain their kinds. Kinds are normally derived from use rather than declared, so most lenses accept anything — which is why this one is rare. |
| Relation “…” isn't declared | The link still draws; it falls back to a plain, untyped edge. That is the right outcome, because a wrong typed edge asserts a specific claim and a plain one only says “connected”. |
…isn't a field this importer stores | An invented frontmatter key — phase, source, confidence. The value is dropped. Put the fact in the body instead. (A vault sync would have kept it, which is what the message means by naming the vault.) |
event_date“…” isn't YYYY-MM-DD | The date is ignored. Note that assisted-generation lanes strip event_date entirely regardless of format — generated content is not permitted to assert dates, however confident it sounds. Dates belong in the prose where a person can check them. |
| Looks like a near-duplicate of “…” earlier in this paste | Two blocks in the same document propose the same subject — Valence Electron and Valence Electrons. Both will import if you let them. Usually you want one, with the other name as an alias. |
exists · skip is not a warning
It means a node with that slug is already in your lens, so this block was left alone. Expected — it is what makes re-pasting the same reply safe, and it is why running a batch twice does not double your graph.
Nothing is overwritten. The existing node keeps its body, its links and any edits you made. If you actually wanted the new version, edit the node directly or change the slug in the paste.
Errors, which do stop a block
| Error | Fix |
|---|---|
| Missing title | No title: and no # heading in the body. Add either. |
| Title has no letters or numbers to form a slug | A title made only of punctuation. Give it a word. |
lens: …isn't a lens you have | Take the suggestion, or create that lens first. |
| Two blocks claim the same slug | Within one paste. Rename one of them. |
| No frontmatter found | The whole paste, not one block. Each node needs a --- block. |
A block with an error is skipped; the rest of the document still imports. Fix it in the box and preview again — nothing is written until you confirm.
Related
- The paste format — every key, and what each does.
- My links didn't connect anything — the warning that costs work.
- Bring your own AI — how to get a batch that produces fewer of these.