refactor(spec)!: the #3896 close-out sweep — fourteen inert authoring keys leave the surface - #4054
Merged
Merged
Conversation
… keys leave the surface The enforce-or-remove worklist across the remaining governed types, each key tombstoned with its prescription (retiredKey) and stripped by a protocol-17 conversion: - action shortcut/bulkEnabled — no keydown path dispatches shortcuts; the multi-select toolbar reads the view's bulkActions. - flow active/template, node outputSchema, errorHandling.fallbackNodeId — active:false never stopped a flow (status is the enforced lifecycle; the default even READ as disabled while the engine treated unset as enabled); faults route via per-node fault edges. - view list responsive/performance, form defaultSort/aria — no renderer read them. List aria/data stay live. form.data SURVIVED: the removal attempt broke the build (defineForm writes data.provider='schema' onto every metadata form; metadata-protocol serves it) — the build refuted the ledger's dead verdict, so the entry is CORRECTED to live instead. The sweep's method note: a ledger verdict is an input to removal, not a substitute for the build's own proof. - dashboard aria/performance + widget performance (+ orphaned PerformanceConfigSchema) — no renderer applied them; widget.performance had no ledger entry and was call-graph-closed fresh across both repos. - agent knowledge (+ orphaned AIKnowledgeSchema) — declaring sources never scoped retrieval (search_knowledge takes sourceIds from the LLM's tool-call args). The same-major topics→sources rename is ABSORBED into the removal pre-release: composed, its effect was unobservable, and the fixture-disjointness contract of the conversion table enforces exactly that judgment. - skill triggerPhrases — never matched; activation is triggerConditions ∩ the agent's skills[] allowlist. Deliberate keeps, recorded so they are not re-litigated: hook label/description and flow description are docs-shaped annotation fields, exempt from enforce-or-remove. Stale report aria/performance ledger entries (schema already clean) deleted as hygiene. Kit: 6 conversions + chain step (chain-replay green); retirement pins across 7 test files; forms pruned (action/flow/dashboard/agent/skill); i18n bundles regenerated (merge mode); baselines edited deliberately (authorable-surface, json-schema.manifest — AIKnowledge/PerformanceConfig unpublished); the surface guard's coverage matcher generalized to ' / '-separated multi-key conversion surfaces (exactly as strict per key); example flows drop template:true (3 sites, caught by the tombstone's tsc leg); SKILL.md + agents.mdx stop teaching the removed keys; v17 notes + upgrade checklist + README ledger rows updated; major changeset. spec 6932 / objectql 1179 / cli 830 green; all spec gates green; examples + downstream-contract typecheck clean. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01QuViRSR1j6GJjf9qGbnqFX
|
The latest updates on your projects. Learn more about Vercel for GitHub. 1 Skipped Deployment
|
github-actions
Bot
added
documentation
Contributor
📓 Docs Drift CheckThis PR changes 3 package(s): 111 hand-written doc(s) reference the affected code and may need an implementation-accuracy re-verification:
|
….md defineSkill example drops triggerPhrases
CI's liveness gate caught the sweep's ledger discipline error: retiredKey
keeps a key in the walked shape, so its entry must STAY (dead + REMOVED
note — the rls.priority precedent), not be deleted (that is the strict-
removal discipline, used for the tool batch where keys left the shape).
All 14 entries restored with verifiedAt + per-key prescriptions; README
counts corrected to match. The local 'pass' before push was a masked exit
code ('| tail -1' swallows the gate's status) — gates now re-run with
explicit exit checks, which also surfaced the SKILL.md defineSkill example
still authoring triggerPhrases (a head-truncated grep miss; the os:check
harness is the authoritative sweeper and is green again).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QuViRSR1j6GJjf9qGbnqFX
…allbackNodeId Found by the corrected sweep after the SKILL.md catch — flows.mdx's defineFlow example authored template: false, and the error-handling section documented fallbackNodeId in both the example and the property table. The :208 'template' hit is the notify node's email template (a different, live config key); the sharing-rule and app 'active' hits are live keys on other types — all three verified before leaving them alone. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01QuViRSR1j6GJjf9qGbnqFX
os-zhuang
marked this pull request as ready for review
July 30, 2026 07:39
baozhoutao
added a commit
that referenced
this pull request
Jul 30, 2026
…he api reference Only conflict was `content/docs/references/api/errors.mdx`, an AUTO-GEN file that both sides touched: main through #4054's authoring-key sweep, this branch through `FieldErrorSchema.label`. Took main's copy and re-ran `gen:schema && gen:docs` rather than hand-merging a generated artifact. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
os-zhuang
added a commit
that referenced
this pull request
Jul 30, 2026
* feat(showcase): demonstrate bulk actions over a selection `bulkActions` — naming declared object actions so each SELECTED record is fanned out through the action runner — had no example in the showcase. It is also the only way to declare a bulk action: `action.bulkEnabled` was retired in spec 17 (#3896 close-out, #4054) precisely because nothing consumed it, and its tombstone prescribes this key. Adds a `bulk_actions` named view on showcase_task — the selection-bar twin of the existing `legacy_row_actions` fixture — naming both an action whose work is a sandboxed script (showcase_mark_done, via the platform action route) and one that is a custom endpoint (showcase_recalc_estimate, whose already-declared `recordIdParam` carries each record's id). Neither is a field patch, which is what distinguishes this from project.view.ts's `bulkActionDefs`: those mass-EDIT through the data API, this dispatches a real action per record. Verified in a browser against this app: selecting all 10 tasks and running each button issues 10 POSTs (one per record, all 200) to /api/v1/actions/showcase_task/showcase_mark_done and /api/v1/showcase/recalc respectively, and the records change server-side. No console errors. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix(showcase): translate the new view label, add the release-nothing changeset The zh-CN bundle must carry every declared label — the i18n coverage ratchet counts untranslated declared strings per example and refused the 456 → 457 growth. And objectstack requires a changeset per PR; an empty one is the sanctioned declaration for a change that releases nothing (the showcase package is private). Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Jack Zhuang <277994282+os-zhuang@users.noreply.github.com> Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The enforce-or-remove worklist across the remaining governed types, inside the v17 window. Every removal got the full retirement kit (retiredKey prescription + protocol-17 conversion + retirement pin tests); the sweep also produced one correction in the opposite direction, which is the most instructive part.
Removed (14 keys, 6 conversions)
shortcut,bulkEnabledbulkActions, never this flagactive,template, nodeoutputSchema,errorHandling.fallbackNodeIdactive: falsenever stopped a flow —statusis the enforced lifecycle, and the default even read as disabled while the engine treated unset as enabled (therls.enabledshape, over-permissive direction);templatehad no reader; node outputs were never validated; faults route via per-node fault edgesresponsive/performance, formdefaultSort/ariaaria/datastay live) — an ARIA claim merely accepted is false accessibility compliancearia,performance, widgetperformancewidget.performancehad no ledger entry and was call-graph-closed fresh across both repos (objectui's virtual scrolling reads the live top-levelvirtualScroll)knowledgesearch_knowledgetakessourceIdsfrom the LLM's own tool-call arguments, never the agent record. A grounding claim in the dangerous directiontriggerPhrasestriggerConditions∩ the agent'sskills[]. The cloud API even served the field back — a dead-end projection that made the false capability look realOrphaned value schemas removed with their keys (
PerformanceConfigSchema,AIKnowledgeSchema, #3950 precedent); the same-majortopics→sourcesrename is absorbed into the knowledge removal pre-release — composed, its effect was unobservable, and the conversion table's fixture-disjointness contract enforces exactly that judgment.The correction:
form.datasurvived, and the build is the refereeThe ledger said form
datawas dead ("no form-path reader in either repo"). The removal attempt broke the build:defineFormwritesdata: { provider: 'schema', schemaId }onto every*.form.tsmetadata form, andmetadata-protocolserves it to the metadata-admin pipeline. So the entry's verdict is corrected to live (with the real evidence and averifiedAt), the conversion narrowed, and the lint tests pin the non-warn. Method note now recorded in the entry: a ledger verdict is an input to removal, not a substitute for the build's own proof.Deliberate keeps (recorded, not skipped)
hook.label/descriptionandflow.descriptionare docs-shaped annotation fields — they document intent for the next reader (per ADR-0033, often a model) even without a runtime consumer. Kept, with ledger notes marking them exempt from enforce-or-remove so future audits don't re-litigate. Stale reportaria/performanceledger entries (schema already clean since the report-liveness close-out) deleted as hygiene.Kit details
spec-changes.json/upgrade guide carry every entry.' / '-separated multi-key conversion surfaces before matching — exactly as strict per key as before; multi-key conversions have been the house style since the tool sweep, and this is the firstretiredKeybatch to exercise that path.template: true(3 sites — two of them caught by the tombstone's tsc leg after the initial grep missed them); SKILL.md andai/agents.mdxstop teaching the removed keys; v17 notes, upgrade checklist, and ledger README rows updated; major changeset.Verification
liveness,empty-state,authorable-surface,api-surface,spec-changes,upgrade-guide,docs251,skill-refs,skill-examples198,i18n).tscclean.Generated by Claude Code