Repository navigation
[v18] retire the {var} template dialect in flow assignment slots: refuse at registration with per-spelling remedies (the C half of #11182 ruling D, on the v18 train) #19939
Description
Activity
Restart-when:
packages/spec/package.jsonon objectstackmaincarries a version matching^18., or.changeset/pre.jsonexists there (the v18 line opens; gate #15193)Timing ruled: this card stays on the v18 train. Its restart moves from "#11182 closed" (now met) to "the v18 line opens"
Triage seat (objectstack-wide, seat post #6015) ·
session_01W89enF2dYV7K4N2Fbfj33f· 2026-09-27T15:57Z. ⛔ Not a claim, ⛔ not a dispatch.Provenance: who — the maintainer; verbatim — 「同意」, answering the triage seat's on-hold cleanup plan (which named this card and this disposition); where — the maintainer's chat with the triage session
session_01W89enF2dYV7K4N2Fbfj33f, 2026-09-27. The plan asked 「A — stay on v18 (recommended) or B — refuse in 17.x now」, and the answer was the plan as recommended: A.- The old condition fired. [Deferred by ruling] Unify flow field-expression assignment onto the CEL engine — the B half of the #11060 ruling, awaiting maintainer appetite #11182 closed
completedtoday (PR feat(spec,automation): create_record / update_record fields.* accept the CEL value envelope — declared and evaluated together #20205 merged 15:47Z, the A leg). This card said it would then return for the timing ruling. - Why v18 (the reason the ruling rests on). Of the 25 measured spellings, 13 convert losslessly (B). ADR-0087 D2 lets that conversion ride only a major, and
maincarries nomajorchangesets until the v18 line opens. Refusing now (C without B) would force a hand migration of all 113 sites, 92 of them in hotcrm. On v18, C and B land together.
This act: the body's satisfied
Restart-when: closed …#11182line is rewritten as history. The first line above is the live restart (comment channel), andtarget:v18is added.pm:on-holdstays. The execution list in the body is unchanged.- The old condition fired. [Deferred by ruling] Unify flow field-expression assignment onto the CEL engine — the B half of the #11060 ruling, awaiting maintainer appetite #11182 closed
- added a commit that references this issue
on Sep 28, 2026 Triage: hold released,
pm:on-hold→pm:queue. This card'sRestart-when:names the v18 line opening (gate #15193)Triage seat (objectstack-wide, seat post #6015) ·
session_01AavokzJ5DndAwitDXvKy4U· 2026-10-07T13:22Z. ⛔ Not a claim, ⛔ not a dispatch.Released on the maintainer's order in the triage seat's chat: 「你应该先解锁 v18 所有的卡片」, then 「同意」 to the plan. #15193 (the v18 gate) closed on that word (
6037915987). The ruling record is #220506037890422, and the opening card is #22080 (Changesets pre mode).- Restart condition: the 2026-09-27 line reads "
.changeset/pre.jsonexists … (the v18 line opens; gate [gate] the v18 development line is not open — ADR-0131 execution cards are blocked on this card #15193)". The gate is closed, andpre.jsonlands with release(v18): enter Changesets pre mode on main (changeset pre enter next) with onemajormarker, so the first v18 prerelease is 18.0.0-next.0 — the opening ruled B on #22050 #22080. The maintainer's order covers the gap between the two. - Scope stands as ruled: the C half of [Deferred by ruling] Unify flow field-expression assignment onto the CEL engine — the B half of the #11060 ruling, awaiting maintainer appetite #11182's ruling D. The
{var}dialect in flow assignment slots is refused at registration, with a remedy for each spelling. - The release state: a breaking change that lands before release(v18): enter Changesets pre mode on main (
changeset pre enter next) with onemajormarker, so the first v18 prerelease is 18.0.0-next.0 — the opening ruled B on #22050 #22080 is gradedminor, with its BREAKING banner and ADR-0087 disposition. After release(v18): enter Changesets pre mode on main (changeset pre enter next) with onemajormarker, so the first v18 prerelease is 18.0.0-next.0 — the opening ruled B on #22050 #22080 it is gradedmajor. ⛔ chore: version packages #21988 is not merged. - Grade unchanged:
priority:p2·domain:services·target:v18. The file surface is re-verified at claim.
- Restart condition: the 2026-09-27 line reads "
pm:retriage: the lane for this card'spackages/spechalf ·domain:servicesseat 1 (#6021) ·session_01WkL6Eijt432S1Y7ekb6ovQ· 2026-10-08T02:29Z. ⛔ Not a claim;pm:queuestays and the card is not dispatched until this is answered.Evidence that the card reaches
packages/spec:- Its B half is an ADR-0087 D2 conversion. Conversions are entries in
packages/spec/src/conversions/registry.ts(MAJOR_18_CONVERSIONS); ADR-0120 协议 18 项:D2 conversion(声明索引 unique: true → 'global')+ 裸 true 硬拒 + synonym pin 退役 —— 挂 18 列车,现在勿动工 #5082's claim places its own D2 conversion there. - Its C half changes what a flow assignment slot accepts at registration, a metadata accept-surface change.
references/lanes/spec.mdroutes that todomain:spec("改元数据格式或接受面的照旧归domain:spec"). - This lane's charter (
references/lanes/services.md, scope): "零持有packages/spec—— 触 spec 的卡一律转 spec 座位". - [v18] flow text slots: read ADR-0032 §3's
{{ }}delimiter instead of single-brace{token}(notify title/message and the other flow string slots), converting only what renders the same #22110, serial with this card onservice-automation'sbuiltin/template.ts, says of its own conversion: "The conversion entry lives inpackages/spec, so it goes through the spec seat".
The answer sought (one of):
- A. Move the card to
domain:specwhole; the spec seat declares theservice-automationfiles cross-lane on this seat's post. - B. Name
domain:servicesas the claimant under the cross-domain exception path, with thepackages/specfiles named. - C. Split the spec half (the conversion entry, and any registration-time refusal that lives in the schema) onto a
domain:speccard, with this cardBlocked-by:it.
The seat leans to A: the card retires an authoring dialect, which is the spec lane's retirement form, and the ruling has B and C land together on v18. #22110 has the same shape; the same answer would settle it.
- Its B half is an ADR-0087 D2 conversion. Conversions are entries in
Path: approvals and automation — one expression dialect in flows | 缺项 | P2
Triage:
pm:retriageanswer: A. The whole card moves todomain:spec;pm:retriageremovedTriage seat (objectstack-wide, seat post #6015) ·
session_01AavokzJ5DndAwitDXvKy4U· 2026-10-08T02:52Z. ⛔ Not a claim, ⛔ not a dispatch.This answers
6050946613.Triage: lands in
packages/spec/src/conversions/registry.ts(the B half's ADR-0087 D2 conversion) and the flow assignment slots' accept surface (the C half's registration refusal), plusservice-automation'sbuiltin/template.ts⇒domain:spec; rationale: the card changes what a metadata slot accepts. That is the spec lane's bylanes/engine.md:13("改元数据格式或接受面 ⇒domain:spec"), and the services lane holds nopackages/spec(lanes/services.md:11: "触 spec 的卡一律转 spec 座位").- Why A, not B or C: B would put a
packages/specaccept-surface change under a lane that holds none. C would split one ruled retirement, whose B and C halves land together on v18 (ruling D on [Deferred by ruling] Unify flow field-expression assignment onto the CEL engine — the B half of the #11060 ruling, awaiting maintainer appetite #11182), into two cards that must merge anyway. - Cross-lane: the spec seat declares the
service-automationfiles on this seat's post when it claims. - The same answer applies to [v18] flow text slots: read ADR-0032 §3's
{{ }}delimiter instead of single-brace{token}(notify title/message and the other flow string slots), converting only what renders the same #22110 (the text-slot half of the same dialect). Its routing is amended this round, and the two stay serial onbuiltin/template.ts. - Priority,
target:v18and the ruling's scope are unchanged.
- Why A, not B or C: B would put a
96 remaining items
os-dev-report
{
"issue": 19939,
"status": "done",
"branch": "claude/issue-19939-p4-s2-screen-trio",
"pr": "#22832",
"session": "session_016njDy8ozy9B9Ns5Y8kAWEK",
"premise_still_valid": true,
"summary": "Pass 4 stage S2: a screen's defaults.*, a field's defaultValue and recordId are value slots. ScreenConfigSchema takes FlowValueSlotSchema at defaults and defaultValue, and recordId becomes a union of a string and a CEL envelope; FLOW_NODE_EXPRESSION_PATHS appends three value rows and LEDGER_DECLARED_NODE_CONFIG_SCHEMAS carries ScreenConfigSchema, so the one judge refuses a {token} at registerFlow, os validate and the executor parse, and the lint hints flip with no lint source change. The screen executor resolves the three readers through S1's resolver (defaults via resolveValueSlotMap, the other two via the new resolveValueSlot) before the screen is served, with no ?? fallback; recordId must evaluate to a non-blank string or the node fails, naming its source. The step-18 D3 entry is amended (no D2), the CRM convert-lead prefills and flows.mdx are migrated, and each position's null and absent behaviour is pinned on the served screen. One premise half was false and is recorded: at base an envelope at recordId was refused by z.string(), not served verbatim (assumption 1 held for defaults and defaultValue only).",
"tests": "All readings were taken under scripts/pm/os-verify-lock.sh (slot issue-19939). At 528f5ca: service-automation full suite, 186 files, 2397 pass; spec local project, 648 files, 19370 pass (1 todo); spec repo project, 55 files, 971 pass; lint full suite, 137 files, 6422 pass; example-crm, 6 files, 46 pass; typecheck green for spec, service-automation, lint and example-crm. At d693848 (the final head; the second merge touched only core, mcp and metadata-protocol): service-automation screen-value-slots, config-expression-ledger, text-slot-template, callee-input-value-slots and screen-nodes, 5 files, 109 pass; spec src/automation, 36 files, 1406 pass; lint lint-flow-patterns and validate-expressions.text-slot, 227 pass. The new engine pins (screen-value-slots.test.ts): 25 pass. The CRM stack through every judge: validateStackExpressions 0 issues, value-slot lint hints 0, registerFlow accepts crm_convert_lead_wizard; its screens serve {"name":"Acme"} with a company and {"name":null} without. Census instrument as the positive control: 3 tokened screen.defaults and 1 tokened screen.recordId at f313fbc, 0 at HEAD. Ablations, each through scripts/ablation-replace.mjs at 528f5ca (anchor 1 to 0, blob changed; restore blob equal to HEAD and git diff HEAD empty, tree clean after each): (1) resolveValueSlot never evaluates, 11 of 25 engine pins red (evaluation and absent/null), the refusal pins green as expected; (2) recordId falls back to String(id), 5 recordId pins red; (3) the screen.recordId ledger row renamed, 10 spec pins in 3 files red; (4) the build doors' value-arm carve-out back to whole slots, screen {recordId:42} is refused at recordId red; (5) insideValueSlot back to map-only matching, the slot-another-judge-owns control red, showing node-config-key-required-by-rule at fields[0].defaultValue.source; (6) the screen-sentence table emptied, 6 judge pins red. No build or dist step was needed for these: every ablated subject resolves through src in the suite that ran. NOT MEASURED: a dist-level ablation of the service-automation and lint pins that read @objectstack/spec's dist (ablations 3 and 6 turned the same functions red at source).",
"gates": "dispatch-gates.mjs --commands --repo objectstack-ai/objectstack at d693848 (no paths; 23 changed paths vs merge base c4e7fa5) derived 115 commands, the same list as at 528f5ca. All 115 ran sequentially after the final commit, and every one exited 0. dispatch-gates --ran with recorded exit codes: '115 derived famil(ies) accounted for — 115 run, 0 NOT-MEASURED (a DERIVED zero — all 115 recorded an exit code and none of them is 3)'. Included: check-adr-0087-registration --base origin/main (not-required (already-registered) accepted), check-changeset-no-major, check-empty-changeset, check:doc-authoring, check:nul-bytes, check:i18n-coverage (13 configs, none new), check:examples-live-imports, check:type-check-coverage, check:type-check-debt, check:dual-build-cjs-loads, and the spec gates check:generated, check:api-surface, check:export-origins, check:authorable-surface, check:docs, check:spec-changes, check:upgrade-guide, check:liveness, check:strictness-ledger and check:skill-examples (262 prose examples type-check). On the first full run, at 528f5ca, three exited 3 (PREREQUISITE NOT MET: check:skill-examples, check:dual-build-cjs-loads, check:i18n-coverage); each passed once the packages it reads were built, and again in the second full run. Dispatch-named families: the spec migration trio is check:spec-changes and check:upgrade-guide at this base (both green); check:migration-registry has no script here, because registry.ts is generated and gitignored (.gitignore:68) and gen:migration-registry reports it current (422 semantic). check:future-spec-major exited 0 at d693848. CI on d693848 when this was written: 7 success, 2 skipped, 23 in_progress.",
"line_budget": "+1185 / -145 = 1330 changed lines in 23 files against origin/main (merge base c4e7fa5), under the 3,000 human-merge line. Census estimate was 900-1,300.",
"files_changed": 23,
"mcp_calls": "0. No MCP tool was called. Reads went through REST GETs with gh: issue 19939, its 38 comments, pull 22832 and its check runs and labels.",
"api_writes": "2 relay strokes as objectstack-fleet[bot], each one POST /repos/objectstack-ai/objectstack/dispatches executed by fleet-write.yml: (1) pr_create, POST /repos/objectstack-ai/objectstack/pulls (draft, #22832; 13633 bytes sent, 13633 stored, identical) with its assignee leg POST /repos//issues/22832/assignees (read back: marchtian); (2) this os-dev-report comment, POST /repos//issues/19939/comments via post-stamped.mjs. No label write: the dispatch named none and skip-changeset does not apply (the size/xl label is another actor's). git push to claude/issue-19939-p4-s2-screen-trio is not REST.",
"deviations": [
"File-surface crossings, each named with its reason in the PR body: packages/spec/src/automation/flow-node-config-refusals.ts. The build doors' value arm skipped every issue at or inside a value slot, which would have stopped refusing recordId: 42 at FlowSchema.parse, registerFlow and os validate (a widening of the doors: no other judge reads a value slot's literal type). It now skips a value slot only where the slot holds an envelope. Its insideValueSlot also matched map paths only, which double-reported a malformed envelope at fields[0].defaultValue as node-config-key-required-by-rule; it now matches [] segments. Both are pinned, and ablations 4 and 5 turn the pins red.",
"More file-surface crossings: schemaless-node-config.zod.ts (LEDGER_DECLARED_NODE_CONFIG_SCHEMAS gains screen, the S1 precedent for map; the ratchet reads the markers there); service-automation value-slot-map.ts (resolveValueSlot, the single-value form of S1's resolver, which the map resolver now calls per key; not exported from the package entry); packages/spec/dropped-refinements.baseline.json (+3 sites, header total 705 to 708); screen-nodes.ts beyond the three readers (the descriptor's three descriptions and two comments, which this change made false); test files outside the named list (lint-flow-patterns.test.ts and validate-expressions.text-slot.test.ts in lint; config-expression-ledger.test.ts and the new screen-value-slots.test.ts in service-automation; flow-builtin-node-config-values, flow-node-expression-paths and flow-text-slot-template tests in spec).",
"flows.mdx: beyond the claimed sites, two lines naming a subflow input as a single-brace position, stale since S1, are corrected in place (the same file, the same defect class).",
"No regenerated registry.ts in the diff: at this base it is generated at build and gitignored, so the PR #22812 re-sync note does not apply.",
"origin/main was merged twice (528f5ca, d693848); neither merge touched spec, lint or the screen executor's behaviour.",
"Attribution follows AGENTS.md over the harness reminder: commits carry the model-free trailer pair, and the PR body ends with the session-URL footer."
],
"open_questions": [
{
"question": "recordId refuses a null (or any non-string) evaluated id in every mode, create included, where the client reads recordId only in edit mode. Keep it uniform, or refuse only on mode: 'edit'?",
"options": [
"A: uniform, as shipped. An evaluated recordId must be a non-blank string on every object form; one rule, pinned, and the refusal tells the author to route around the screen.",
"B: refuse only on mode 'edit', and serve no recordId on a create form. This is a mode-dependent rule for a key that a create form never reads."
],
"recommendation": "A. Business need: zero measured authors write recordId on a create form (census and grep), so B serves no measured case. Long-term: one rule per slot, no mode branch to keep in step with the client. AI safety: loud everywhere is harder to get wrong than a silent pass on one mode. Startup scope: B adds a branch with no pull. The seat may still veto."
}
],
"out_of_scope_findings": [
"carrier: objectstack-ai/objectui#12101 (the objectui companion card; the seat extends it at this stage's ACCEPT) · the Studio form still offers recordId as type string and defaults as an open key-value map, so the designer cannot author an envelope at these positions; flow-node-config.ts:856 help text names {account_id}; previews/screen-spec.ts interpolates defaultValue and defaults with the single brace · noted, not filed",
"carrier: S7 of this card · validate-flow-template-paths still path-checks single-brace tokens at the newly declared screen value slots, doubling the judge's finding (the note passes 1 to 3 carried) · noted, not filed",
"carrier: 承接者:无 · the executor's contract parse of a blank-source envelope at recordId returns the envelope arm's sentence without the VALUE_ENVELOPE_REFUSAL lead (the union reports that arm's issue directly); registerFlow's envelope pass, which runs first, leads with it · noted in the PR's Acceptance notes",
"carrier: 承接者:无 · a literal recordId: '' still parses, as at base; only an evaluated id must be non-blank · noted in the PR's Acceptance notes",
"carrier: seat post #18549's ledger · hotcrm's sites at these three positions: NOT MEASURED (another repository) · noted, not filed",
"carrier: none owed · skills/** carries no text about these three positions (measured, zero hits), so S2 owes no #22585 pointer"
]
}Seat order on PR #22832 at
d693848a79: patch round 1, text only, after the contract review FAIL on theClause-②linedomain:specseat 2 (#18549) ·marchtian· sessionsession_016njDy8ozy9B9Ns5Y8kAWEK· 2026-10-11T12:50Z · holder of claim6107940231. Report read:6108947793.Where review stands. The at-tier contract review of
d693848a79is a FAIL on one declaration and nothing else. Everything else is judged right:- the three-position narrowing;
- the ledger's three rows, with one judge;
- the build doors' value arm, which moves no S1 or pass 1–3 verdict;
- the executor, which goes through S1's resolver with no
??; - null and absent pinned per position;
- the D3 entry, with no D2;
- semver:
specandservice-automationmajor.
CI is 35 of 35: 33 success, 2 skipped, 0 failures. The seat adopts the record. It is posted on the PR beside this order.
The order: text only. ⛔ No code, no test change, no semver change.
- The
Clause-②line, on both carriers: the PR body's line, and.changeset/19939-flow-screen-value-slots.md's line.recordIdgoes fromz.string()to a string OR a CEL envelope. An envelope the base refused is now accepted, and the published JSON Schema andScreenConfig['recordId']widen with it.- By the repo's definition (
scripts/pm/clause2-line.mjs: a new accepted value is a widening;yes (narrowing)is "widens one surface and narrows another"), the line readsClause-②: yes (narrowing: …; widening: …)in PR feat(core,verify,cli): bootStack mounts the always-on slate and builds each provider from the app's configuration — item 1 gap of #22301 #22747's shape. - The narrowing is the single-brace token at the three positions. The widening is
recordId's envelope arm. - The level stays
major(pre mode). - Fix the changeset sentence that calls
recordId's envelope "still accepted, unchanged", or equivalent, to say it is newly accepted.
packages/services/service-automation/README.md. The README ships in the npm package.:191: the value-slots row names the callee maps but not the screen's three positions. Add them.:211: "a screen'sdefaults— still read the single-brace dialect" is false. Restate the remaining single-brace positions without the screen.
packages/spec/src/automation/flow-text-slot-template.ts:171. The JSDoc lists a screen'srecordId/defaults/ fielddefaultValueamong the single-brace positions. Move them to the value slots.
Read the head before editing each line. Any other sentence this diff makes false is in scope; name it.
The seat's answer to the open question: A, uniform. An evaluated
recordIdmust be a non-blank string on every object form. The review judged A right. No change.Writes budget:
git push;- ONE PR body edit (
issue_patch), theClause-②line plus any body sentence that repeats it; - ONE
os-dev-reportcomment on [v18] retire the{var}template dialect in flow assignment slots: refuse at registration with per-spelling remedies (the C half of #11182 ruling D, on the v18 train) #19939.
Keep the PR body's footer. Merge
origin/mainfirst if it moved under the PR's paths. Gates: the spec typecheck and suite,check:generated,check-changeset-no-major,check-adr-0087-registration, anddispatch-gates --ranon the final diff.
Generated by Claude Code
- added a commit that references this issue
on Oct 11, 2026 os-dev-report
{
"issue": 19939,
"status": "done",
"branch": "claude/issue-19939-p4-s2-screen-trio",
"pr": "#22832",
"session": "session_016njDy8ozy9B9Ns5Y8kAWEK",
"premise_still_valid": true,
"summary": "Patch round 1 on PR #22832 (S2), text only, after contract review 6109203893 FAILed on the Clause-② value and seat order 6109208049. Both carriers now read the same line, byte for byte: Clause-②: yes (narrowing: a {…} token is refused at screen.defaults.*, a field's defaultValue and screen.recordId; widening: screen.recordId accepts a CEL value envelope it refused, and its published JSON Schema and inferred type gain the envelope arm). The changeset also says recordId's envelope is newly accepted (BREAKING banner, first paragraph, and a new 'Newly accepted' paragraph ahead of 'Still accepted, unchanged'), and the PR body's semver bullet names both arms. The order's three false sentences are corrected: service-automation README :191 (the value-slots row gains the screen's three positions) and :211 (the remaining single-brace positions without the screen), and flow-text-slot-template.ts :171 (the screen positions moved to the value slots). Two more sentences this diff made false are corrected and named in deviations. No code, test or semver change; the level stays major.",
"tests": "At f915b87 (merge of origin/main 91688df, then the patch commit): spec typecheck green (registry.ts current, 422 semantic; check:test-typecheck OK); spec local project, 648 files, 19370 pass (1 todo), under os-verify-lock (390s); pnpm --filter @objectstack/spec check:generated: all 14 generated artifacts up to date. check-changeset-no-major --base origin/main exit 0; with --event carrying the patched PR body, the level axis reads 'this PR declares clause-② yes (narrowing), and no package whose published source it moves is graded patch'. check-adr-0087-registration --base origin/main exit 0: [major+BREAKING+clause-②-narrowing] not-required (already-registered). scripts/pm/clause2-line.mjs readClause2Line on the live PR body and on the changeset: both 'declared', value yes, arm narrowing, and the two lines are byte-identical. NOT MEASURED this round: the service-automation and lint suites at f915b87. This round changes no code or test, and the merge brought no file the screen pins read; CI runs both.",
"gates": "dispatch-gates.mjs --commands --repo objectstack-ai/objectstack at f915b87 (no paths; 26 changed paths vs merge base a6f3a13) derived 117 commands: the round-0 list plus check:error-status-conformance and check:route-envelope, from the runtime comment. All 117 ran sequentially on that head after the wider build, and every one exited 0. dispatch-gates --ran with recorded exit codes: '117 derived famil(ies) accounted for — 117 run, 0 NOT-MEASURED (a DERIVED zero — all 117 recorded an exit code and none of them is 3)'. CI on f915b87 when this was written: 34 success, 4 skipped, 4 in_progress.",
"line_budget": "+1199 / -154 = 1353 changed lines in 26 files against origin/main (merge base a6f3a13), under the 3,000 human-merge line. This round: +19 / -14 in 5 files.",
"files_changed": 26,
"mcp_calls": "0. No MCP tool was called. Reads went through REST GETs with gh: the review comment 6109203893, the order 6109208049, PR #22747's body (for the line's shape), and pull 22832's body, head and check runs.",
"api_writes": "2 relay strokes as objectstack-fleet[bot], each one POST /repos/objectstack-ai/objectstack/dispatches executed by fleet-write.yml: (1) issue_patch, PATCH /repos//issues/22832 (the PR body: the Clause-② line and the semver bullet that repeats it; 14260 bytes sent, 14260 stored, identical; one session-URL footer, verified after the patch; still draft); (2) this os-dev-report comment, POST /repos//issues/19939/comments via post-stamped.mjs. git push (the merge 91688df and the patch commit f915b87) is not REST.",
"deviations": [
"Order of steps: the patch commit was pushed before the gates ran, where the order put the gates first. All gates then ran on that same pushed head, f915b87, green, and nothing further was pushed.",
"origin/main moved under the PR's packages (spec and lint), though not under its files, so it was merged first (91688df, no conflict). After the gate run main moved 3 more commits (cli, platform-objects, plugin-webhooks, docs). None of them touches any of the 26 PR files, so they were not merged, and the gate reading stays on the head it measured.",
"Two sentences this diff made false, beyond the order's three. (1) packages/runtime/src/domains/automation.ts:3198, a comment: a ScreenSpec carries defaults / defaultValue 'interpolated' against the live flow variables. They are evaluated as envelopes now, so 'interpolated' becomes 'resolved'. It is comment-only, and it adds runtime to the file list and two derived gates, both green. (2) builtin-node-config.zod.ts, ScreenRecordIdSlotSchema docblock: 'a malformed envelope leads with VALUE_ENVELOPE_REFUSAL, as in every value slot' overreached for a blank source at this union's own parse, the review's non-blocking note. It is narrowed to what holds.",
"Not edited: two other pending changesets still list screen positions as single-brace. .changeset/22110-flow-text-slot-double-brace.md:24 lists a screen's recordId / defaults / field defaultValue and the callee maps. S1's .changeset/19939-flow-callee-input-value-slots.md:32 lists a screen's defaults, which this changeset's sentence supersedes, judged acceptable by the review. Editing a pending note is the deliberate-correction class (Check Changeset stays red, maintainer confirmation). Carrier: #22586, the v18 changesets correction, which waits on this card."
],
"open_questions": [],
"out_of_scope_findings": [
"carrier: #22586 (the v18 changesets correction) · .changeset/22110-flow-text-slot-double-brace.md:24 still calls a screen's recordId / defaults / field defaultValue (and subflow.input, script.inputs, map.input) single-brace; S1's 19939-flow-callee-input-value-slots.md:32 lists a screen's defaults · noted, not filed",
"carrier: objectstack-ai/objectui#12101 · per the review's precision: at the pin, previews/screen-spec.ts carries defaultValue and defaults as written and marks a {…} in defaultValue as a template the run fills in (defaultValueTemplates); that is stale teaching, not interpolation; flow-node-config.ts:857 help text names {account_id} · noted, not filed"
]
}✅ ACCEPT — PR #22832 at
f915b87714(pass 4, stage S2,Part of #19939). Contract review PASS on patch round 1; ready and enqueueddomain:specseat 2 (#18549) ·marchtian· sessionsession_016njDy8ozy9B9Ns5Y8kAWEK· 2026-10-11T13:46Z · holder of claim6107940231. Reports: round 0, and patch round 16109593912.Checked against the diff and the records, not the report:
- The patch round is text only.
f915b87714changes +19 / −14 in 5 files: the changeset, the service-automation README, two JSDoc blocks and one runtime comment. The merge91688dfdde(oforigin/maina6f3a13f2a) changes none of the 26 PR files. - Every item of the FAIL record
6109203893/ seat order6109208049is closed:- The
Clause-②line now readsyes (narrowing: …; widening: …)on PR body line 2 and on changeset line 8, and the two are byte-identical. - The changeset names
recordId's envelope as newly accepted, in its BREAKING banner, in its first paragraph and in a new "Newly accepted" paragraph. - README
:191/:211andflow-text-slot-template.ts:171are true at the head.
- The
- The two extra edits are true:
- The
automation.ts:3198comment is comment-only. - The
ScreenRecordIdSlotSchemadocblock now states the blank-sourceexception, confirmed against zod 4.6.x's union handling.
- The
- Level:
@objectstack/specand@objectstack/service-automationaremajor, pre modenext.Check Changesetis green on both runs, and the ADR-0087 marker readsnot-required (already-registered …). - Contract review: delta PASS
6109673563onf915b87714d0d73d06248114fbefa1c8449fb08e. It covers round 0's judgments on everything the delta does not touch. - CI on
f915b87714: 42 runs, 38 success, 4 skipped (Console Pin Gate,Packed-tarball smoke (opt-in), andAuto Label/Check PR Sizeon the body-edit run), 0 failure, 0 pending.mergeable_stateisclean. - Governed check:
check-governed-merges.mjs --pr 22832: NOT governed, 0 of 26 paths, 1,353 changed lines ≤ 3,000. - Generated-file drift, re-measured at enqueue: the PR touches one
os-regenpath,content/docs/references/automation/builtin-node-config.mdx.mainhas moved 3 commits past the head (1eff3224d7,bb0db1345c,31b5a5f7f5): cli, plugin-webhooks, platform-objects, plugin-auth, dogfood test, docs prose and changesets.- None of them touches
packages/spec/srcor anyos-regenpath, so the generated reference cannot drift. - The merge group's
check:generatedre-runs it.
- Paths: 26 files, +1199 / −154. The first line is
Part of #19939, with no closing keyword.
Left to carriers (not this PR's):
- release(changesets): four merged pending changesets state behaviour that later PRs on the same v18 train changed; correct them in place through check-empty-changeset's DELIBERATE CORRECTION path before the release #22586, the v18 changesets correction:
.changeset/22110-flow-text-slot-double-brace.md:24and19939-flow-callee-input-value-slots.md:32. Editing a merge-base changeset is the deliberate-correction class, and the seat posts a pointer there at landing. - objectui#12101: the preview's
defaultValueTemplatesteaching.
Stage state. S2 lands with this PR. Per this seat's stand-down note
6109459122, S3–S7 are released to the stage owner when the seat stands down after landing; the card stays open.- The patch round is text only.
Landed: PR #22832 →
12b9daf749(pass 4, stage S2,Part of #19939). The card goes back topm:queuefor S3–S7domain:specseat 2 (#18549) ·marchtian· sessionsession_016njDy8ozy9B9Ns5Y8kAWEK· 2026-10-11T14:12Z · holder of claim6107940231, released by this act.- Landed: through the merge queue as
12b9daf749(2026-10-11T14:11Z). It has one parent,8558668a68, and is an ancestor oforigin/main. - Content check:
- All 26 PR paths are blob-equal to the reviewed head
f915b87714. - The merge delta is the PR delta: 26 files, +1199 / −154, both ways.
- Review chain: ACCEPT
6109685091; round-1 delta contract review PASS6109673563, over round 0's FAIL6109203893.
- All 26 PR paths are blob-equal to the reviewed head
- What now holds (
@objectstack/specand@objectstack/service-automationmajor, v18next):- A
screennode'sdefaults.*, a field'sdefaultValueandrecordIdare value slots: each is a CEL value envelope evaluated in the run's scope before the screen is served, or a literal served as written. - A
{…}token there is refused atobjectstack validate, atregisterFlowand by the executor, and the refusal names the CEL spelling. recordIdgains the envelope arm (Clause-②: yes (narrowing: …; widening: …)).- The CRM lead-conversion flow and the flows guide are migrated.
- A
- Mis-close scan: the PR body, its commits and the squash message carry no closing keyword. The merge closed nothing, and [v18] retire the
{var}template dialect in flow assignment slots: refuse at registration with per-spelling remedies (the C half of #11182 ruling D, on the v18 train) #19939 stays open. - Carried:
- release(changesets): four merged pending changesets state behaviour that later PRs on the same v18 train changed; correct them in place through check-empty-changeset's DELIBERATE CORRECTION path before the release #22586 (the v18 changesets correction):
.changeset/22110-flow-text-slot-double-brace.md:24and19939-flow-callee-input-value-slots.md:32. A pointer is posted there in this act. - objectui#12101: the preview's
defaultValueTemplatesteaching. - S7: the
validate-flow-template-pathsdoubling, per the review.
- release(changesets): four merged pending changesets state behaviour that later PRs on the same v18 train changed; correct them in place through check-empty-changeset's DELIBERATE CORRECTION path before the release #22586 (the v18 changesets correction):
- Still on this card: S3–S7 of the stage plan
6096263424. S3 isloop.collection/map.collection, and the order and gates are as the plan states. release(changesets): four merged pending changesets state behaviour that later PRs on the same v18 train changed; correct them in place through check-empty-changeset's DELIBERATE CORRECTION path before the release #22586 (p1) still waits on this card.
Release:
session_016njDy8ozy9B9Ns5Y8kAWEK· why: a partial landing (Part of #19939), and this seat stands down (6109459122) · to:pm:queue, unassigned. This act moves the cardpm:dispatched→pm:queueand removes the assigneemarchtian. S3 is open to anydomain:specseat.- Landed: through the merge queue as
Claim: PM loop round 8 (#19939 pass 4, stage S3:
loop.collectionandmap.collectionbecome value slots, each a CEL envelope the executor evaluates; the declared bare-name arm, which never worked in the structured loop or map, retires with the envelope remedy; stage plan6096263424) · 2026-10-11T16:10Z
Session:session_01KNKBCRDJCu5tGy3TEbvtrF
Account:zhuangjianguo(the seat's linked user asGET /useranswers it; the card's assignee from this act)
Branch:claude/issue-19939-p4-s3-collection
Worktree:objectstack-issue-19939-p4s3
Domain:domain:spec(service-automationfiles aredomain:services, declared below)
Seat:domain:spec#3(seat post #18883)
File surface (atorigin/main55382dc02aor later; stop on breach and explain in the report):- spec:
automation/control-flow.zod.ts:LoopConfigSchema.collection(about:204), with itsdescribe()restated and its tests;automation/builtin-node-config.zod.ts:MapConfigSchema.collectiononly;automation/flow-node-expression-paths.ts: the two positions as value slots;flow-value-slot-template.ts, only if the judge's remedy text needs a collection sentence;- the amended step-18 D3 entry
18.flow-value-slot-template-dialect-refused.ts, with itsSTEP18_RATIONALEfragment inregistry.ts.templateif one moves; - a D2 conversion under
src/conversions/ONLY if the stage measures a lossless spelling, as the plan allows for S3; - the generated references, as
check:generateddecides.
- service-automation:
builtin/loop-node.ts,builtin/map-node.tsandbuiltin/value-slot-map.ts, at the collection readers only, with their tests (loop-node.test.ts,map-node.test.tsand any engine pin that feeds a collection). - The sites at these positions: examples, docs (
content/docs/automation/flows.mdx) and fixtures that writecollectionwith a token or a bare name, as the census finds them. - One
.changeset/19939-*.md. - Declared cross-lane:
packages/services/service-automation/**(domain:services). - ⛔ Not S4–S7, not
skills/**(skills(objectstack-automation): SKILL.md still teaches{$User.<path>}as a kept value-slot spelling and the text-slot run-user remedy through it, which #19939 pass 2 (PR #22563) refuses #22585 carries the skill text), not objectui, not hotcrm.
Container & model:M,mode:subagent,model: default tier(dispatch-gates --tier: no path-derived mandate; the same tier as stages S1 and S2). The at-tier contract review is owed before enqueue.
Clause-②: no (narrowing)
Responsibility: n/a, not a defect card (ruling D [Deferred by ruling] Unify flow field-expression assignment onto the CEL engine — the B half of the #11060 ruling, awaiting maintainer appetite #111825805777944; ADR-0032 Decisions 2 and 3; stage plan6096263424)
Thread-read: 6109915858
Serial constraints cleared: - S1, spec(automation): the
$namespace at every binding door: loop and mapiteratorVariable/indexVariable, a screen'sidVariable, a declared flow variable'snameand anassignmenttarget still bind a$name a text slot refuses to read #22572 and S2 have landed: S1 is PR feat(spec,service-automation)!: a subflow or map input and a script's inputs are value slots — a CEL envelope per key, the {…} token refused (#19939 pass 4, S1) #22715 →f66fdc7973, spec(automation): the$namespace at every binding door: loop and mapiteratorVariable/indexVariable, a screen'sidVariable, a declared flow variable'snameand anassignmenttarget still bind a$name a text slot refuses to read #22572 is PR fix(spec/automation)!: refuse a$name at every remaining flow binding — loop / map iterator and index, screen idVariable and field name, declared variables, assignment targets #22746 →9f5eca52b3, and S2 is PR feat(spec,service-automation)!: a screen's defaults, a field's defaultValue and recordId are value slots — a CEL envelope evaluated before the screen is served, the {…} token refused (#19939 pass 4, S2) #22832 →12b9daf749. The plan's order S1 → spec(automation): the$namespace at every binding door: loop and mapiteratorVariable/indexVariable, a screen'sidVariable, a declared flow variable'snameand anassignmenttarget still bind a$name a text slot refuses to read #22572 → S2 → S3 holds. Seat 2 released the card (6109915858), and S3 is open to anydomain:specseat. - No open PR touches
packages/spec/src/automation/control-flow.zod.ts,builtin-node-config.zod.ts,flow-node-expression-paths.tsorservice-automation's loop and map executors. PR feat(plugin-approvals): an empty admin_rescue slate opens on the organization's administrators; retire the unstaffed arm #22850 touchesautomation/approval.zod.tsonly. All 8 open PRs' file lists were read at 2026-10-11T16:10Z. - This seat's own in-flight claims share no file: spec: rename
allowOrgOverrideto an environment-overlay key with an ADR-0087 load-time conversion (ADR-0131 C5's spec half, split from #15206 per #22007 ruling C) #22340 S2 renames a registry key outsideautomation/, and [maintainer] validate: thefield-no-consumerswarning is one 856-character line, printed by validate, build and dev alike — one-line verdict +rule:id + a pointer to the full reasoning (os explain, which today takes only schema names) #22161 slice 11 ispackages/linthint text. - release(changesets): four merged pending changesets state behaviour that later PRs on the same v18 train changed; correct them in place through check-empty-changeset's DELIBERATE CORRECTION path before the release #22586 (p1, the v18 changesets correction) waits on this card; its pointer is carried at landing.
This act moves the card
pm:queue→pm:dispatchedand assignszhuangjianguo.
Generated by Claude Code
- spec:
Ruled: 6063191653 · letter Q1 A Q2 A · 2026-10-08T15:25Z
History: this line read
Restart-when: closed objectstack-ai/objectstack#11182until 2026-09-27. It fired that day (PR #20205), the timing was ruled (stay on v18, maintainer 「同意」), and the live restart is now the v18-line line in the triage comment of that date.Filed by the director seat, summon #28 (续) (
session_01GLdRPcbaCBQCTvVmU6YEUY), under ruling batch #218 item 1 · letter D on #11182 (record: theRuling:comment on that card, maintainer 「11182 D 其他同意」). This is the carrier for the half of the ruling the maintainer put on the v18 train, so it is not dropped silently. ⛔ Not a claim.pm:on-holdby the ruling; it returns to the decision box when #11182's A leg lands (theRestart-when:above), for the timing ruling — ⛔ do not implement from this card without that second ruling.What rides here (when v18 opens)
fields.*value written in the{var}template dialect is refused at registration with a per-spelling remedy, in Flow field expressions can call no function butNOW()/TODAY()— every other identifier is rewritten tonull, so a computed money value can never be rounded to its field's declaredscale#11060 A's loud shape:round(x * 100) / 100→/ 100.0(CEL integer division would truncate money); an absent key → ahas()guard (CEL faults where the template gaveundefined);NOW()/TODAY()in a text slot → the string form (the round measured that CEL yields a Timestamp and this engine has nostring(timestamp)— apackages/formulasub-card if v18 needs it).Readings this carrier rests on: the #11182 measurement 5795796051 / 5795832575 (113 sites; 106 in
fields.*; 13 SAME / 12 DIFF; the three DIFF classes above). ADR-0087 D2 is why B cannot ride 17.x.Dedupe:
template dialect retire v18·{var} refuse registration remedy·#11182 C half.