Reproduction
The V2 command implementation on current main (ea544d88405590eb91b5132eb73ef28546b6b0f3) restricts dataset create/infer/chain-ingest types to user_event|multi_modal. The generated root help reflects that restriction, but several advertised examples still use values rejected by the same path:
docs/COMMANDS.md advertises vs dataset create --type <item|event|behavior|image_text|video|user-event|document>.- Root help examples show
vs dataset ingest --file ... --type item.
- All 13 localized READMEs describe
vs dataset create ... --type item as the manual fallback.
- V2 acceptance fixtures/tests retain legacy
item, video, and document assumptions and fail against the current validation/render behavior.
This drift follows the V2 onboarding migration in #36 and later command changes. I searched all-state issues, all PR titles/bodies, and repository history for the affected dataset types and found no existing report or pending fix.
Expected behavior
The public V2 command docs, examples, and acceptance expectations should use the types the current implementation accepts. If legacy item/video/document flows remain intentionally supported elsewhere, the docs should clearly separate those commands from the V2 create/infer/chain-ingest path.
Reproduction
The V2 command implementation on current
main(ea544d88405590eb91b5132eb73ef28546b6b0f3) restricts dataset create/infer/chain-ingest types touser_event|multi_modal. The generated root help reflects that restriction, but several advertised examples still use values rejected by the same path:docs/COMMANDS.mdadvertisesvs dataset create --type <item|event|behavior|image_text|video|user-event|document>.vs dataset ingest --file ... --type item.vs dataset create ... --type itemas the manual fallback.item,video, anddocumentassumptions and fail against the current validation/render behavior.This drift follows the V2 onboarding migration in #36 and later command changes. I searched all-state issues, all PR titles/bodies, and repository history for the affected dataset types and found no existing report or pending fix.
Expected behavior
The public V2 command docs, examples, and acceptance expectations should use the types the current implementation accepts. If legacy item/video/document flows remain intentionally supported elsewhere, the docs should clearly separate those commands from the V2 create/infer/chain-ingest path.