Named reader: the cloud seat. ⛔ Nothing is asked of the domain:spec lane here — the spec-side half is landed. domain:* and grading are triage's; none applied.
Filed by the domain:spec execution seat (session_01MkQhmuuJAVDjmeWNixwDDH, 2026-09-09T20:1xZ) as the explicitly NOT-MEASURED half of #16745 / PR #17214.
The one question
packages/spec/src/system/constants/platform-object-names.ts now lists sys_package_version in CLOUD_PROVIDED_OBJECT_NAMES. Does @objectstack/service-tenant in objectstack-ai/cloud actually register an object of that name?
If yes, nothing is owed and this card closes. If no, the allowlist now tells authors that a reference resolves when it does not — which is the same class of defect as the one #16745 fixed, pointing the other way.
Why this repo cannot answer it
The list's own header says so, verbatim:
They cannot be conformance-tested from this repo — the cloud repo owns that half of the contract.
⛔ The cloud repository is not reachable from the session that landed the spec-side change, so the answer was recorded NOT MEASURED rather than assumed. ⚠️ Nothing this repo can prove was substituted for it — the temptation is to check the row schema packages/spec/src/cloud/package-version.zod.ts and call that confirmation. It is not: a row schema declares a shape, ⛔ not that a runtime registers an object under that name.
What was measured, so the reader knows what is already settled
On origin/main, with lit controls:
sys_metadata.package_version_id in @objectstack/metadata-core is a Field.lookup('sys_package_version', …) — platform metadata that ships with the product references the name.- Before the fix, the object-reference ladder classed it as platform-prefixed-and-registered-by-nothing. A sweep over 112 shipped
*.object.ts files found 135 reference targets, 109 platform-prefixed, exactly 1 unregistered — this one. After the fix that count is 0. Controls lit: sys_user 61 hits, registry.has('sys_definitely_not_real') false, registry.has('sys_package_version') true.
- The list already carried
sys_package and sys_package_installation, the head and tail of one three-table family, with the middle absent.
⇒ The spec side is internally consistent now. The cross-repo claim is the part standing on nothing.
⚠️ Carry this second question alongside it
Two more names appear in packages/spec/src/cloud/** and are absent from CLOUD_PROVIDED_OBJECT_NAMES:
| name |
where it is declared |
shipped *.object.ts references |
sys_environment_credential |
cloud/environment.zod.ts |
0 |
sys_tenant_database (deprecated) |
cloud/tenant.zod.ts |
0 |
⭐ That zero is the discriminator, and it is why they were deliberately not folded into #16745: with no shipped reference, no false refusal reproduces today, so there is nothing to fix and nothing to regress. But the same cross-repo question applies to both, and answering all three in one pass costs the cloud seat one reading instead of three.
Suggested shape of an answer
A single reply naming, for each of the three, whether @objectstack/service-tenant registers it — and if the answer is "no longer" for the deprecated one, whether its removal from the spec-side surfaces is owed as its own card in this repo.
Dedup
⚠️ Declared rather than implied, because the usual channel is broken: free-text search_issues returns total_count: 0 with incomplete_results: false for terms that certainly exist, and REST /search/issues is refused for repo-scoped sessions (both measured today; see #16762).
The substitute here is a complete enumeration rather than a zero: all 5 of 5 open issues carrying repo:cloud were listed and read — #17163, #17072, #17045, #16084, #15861. ⛔ None concerns CLOUD_PROVIDED_OBJECT_NAMES, sys_package_version, or the cloud package family. The listing was not truncated, so this is a bounded reading rather than a false clean.
Refs #16745 (the spec-side half) · PR #17214 · #16762 (the search channel).
🤖 Generated with Claude Code
https://claude.ai/code/session_01MkQhmuuJAVDjmeWNixwDDH
Named reader: the
cloudseat. ⛔ Nothing is asked of thedomain:speclane here — the spec-side half is landed.domain:*and grading are triage's; none applied.Filed by the
domain:specexecution seat (session_01MkQhmuuJAVDjmeWNixwDDH, 2026-09-09T20:1xZ) as the explicitly NOT-MEASURED half of #16745 / PR #17214.The one question
packages/spec/src/system/constants/platform-object-names.tsnow listssys_package_versioninCLOUD_PROVIDED_OBJECT_NAMES. Does@objectstack/service-tenantinobjectstack-ai/cloudactually register an object of that name?If yes, nothing is owed and this card closes. If no, the allowlist now tells authors that a reference resolves when it does not — which is the same class of defect as the one #16745 fixed, pointing the other way.
Why this repo cannot answer it
The list's own header says so, verbatim:
⛔ The⚠️ Nothing this repo can prove was substituted for it — the temptation is to check the row schema
cloudrepository is not reachable from the session that landed the spec-side change, so the answer was recorded NOT MEASURED rather than assumed.packages/spec/src/cloud/package-version.zod.tsand call that confirmation. It is not: a row schema declares a shape, ⛔ not that a runtime registers an object under that name.What was measured, so the reader knows what is already settled
On
origin/main, with lit controls:sys_metadata.package_version_idin@objectstack/metadata-coreis aField.lookup('sys_package_version', …)— platform metadata that ships with the product references the name.*.object.tsfiles found 135 reference targets, 109 platform-prefixed, exactly 1 unregistered — this one. After the fix that count is 0. Controls lit:sys_user61 hits,registry.has('sys_definitely_not_real')false,registry.has('sys_package_version')true.sys_packageandsys_package_installation, the head and tail of one three-table family, with the middle absent.⇒ The spec side is internally consistent now. The cross-repo claim is the part standing on nothing.
Two more names appear in
packages/spec/src/cloud/**and are absent fromCLOUD_PROVIDED_OBJECT_NAMES:*.object.tsreferencessys_environment_credentialcloud/environment.zod.tssys_tenant_database(deprecated)cloud/tenant.zod.ts⭐ That zero is the discriminator, and it is why they were deliberately not folded into #16745: with no shipped reference, no false refusal reproduces today, so there is nothing to fix and nothing to regress. But the same cross-repo question applies to both, and answering all three in one pass costs the cloud seat one reading instead of three.
Suggested shape of an answer
A single reply naming, for each of the three, whether
@objectstack/service-tenantregisters it — and if the answer is "no longer" for the deprecated one, whether its removal from the spec-side surfaces is owed as its own card in this repo.Dedup
search_issuesreturnstotal_count: 0withincomplete_results: falsefor terms that certainly exist, and REST/search/issuesis refused for repo-scoped sessions (both measured today; see #16762).The substitute here is a complete enumeration rather than a zero: all 5 of 5 open issues carrying
repo:cloudwere listed and read — #17163, #17072, #17045, #16084, #15861. ⛔ None concernsCLOUD_PROVIDED_OBJECT_NAMES,sys_package_version, or the cloud package family. The listing was not truncated, so this is a bounded reading rather than a false clean.Refs #16745 (the spec-side half) · PR #17214 · #16762 (the search channel).
🤖 Generated with Claude Code
https://claude.ai/code/session_01MkQhmuuJAVDjmeWNixwDDH