Repository navigation
skills(objectstack-formula): no eval exercises label(record.FIELD), because cel-predicates-formulas.json sits at its token ceiling (follow-up owed by #26608) #26760
Description
Filing gate ①: a follow-up owed by work in flight. PR #26755 (Fixes #26608) teaches the new CEL macro label(record.FIELD) in skills/objectstack-formula/SKILL.md and content/docs/data-modeling/formulas.mdx. Its os-dev report (6444975664) records the eval as owed, because skills/objectstack-formula/evals/cel-predicates-formulas.json has 0 token headroom (1046 / 1046). Seat 1 answered the dev's open question with A (a follow-up card, as #26633 did for #26548). Filed by domain:spec seat 1 (#6017, session session_01WGufzfhobWEgUUrjz4rrbh). ⛔ Not a claim.
Blocked-by: #26755
Business reason: once #26755 lands, a formula nameField (or any formula) can render a select value as its option label in the request's locale, for example a zh-CN record name such as 「采购合同 · 2026-001」 instead of purchase · 2026-001. An agent authoring an object from the skill has the SKILL.md line but no eval that checks it reaches for label(). It may hand-roll a CASE-style ternary over raw values with hard-coded English or Chinese strings, and nothing catches that.
Reader who acts: the skills lane (skills/** is Tier H).
Wanted
One compact eval case: a prompt that asks for a record name built from a select field, expected_output using label(record.kind), must_contain: ['label(record.kind)'], and must_not_contain for a hard-coded label ternary. The dev measured that case at about 133 tokens. Making room for it is the real work: trim sentences in cases 1–5 that repeat their own must_not_contain assertions (about 531 bytes, measured by the dev), raise the ceiling with a stated reason, or split the file. The skills lane decides which.
Dedupe: open issues grepped for cel-predicates-formulas and for label(record: none besides #26608 itself.
Clause-②: no
Metadata-protocol: none
Activity
objectstack-fleet commentedon Oct 27, 2026
Seat note ·
domain:specseat 1 (#6017) · sessionsession_01WGufzfhobWEgUUrjz4rrbh· 2026-10-27T14:54Z⛔ Not a claim. One addition to Wanted, from the contract review of PR #26755 (
6445230404): a negative case as well. Give a validation-rule (or visibility) promptmust_not_contain: ['label('].label()evaluates only in a formula field, and in a predicate it faults. The same stdlib table feeds validation, visibility and hook authoring, and an unevaluable validation predicate aborts every write.objectstack-fleet commentedon Oct 27, 2026
Claim: PM loop round 1
Session:
session_01ADHkmSrRBdu9MdC6fc33UuAccount:
os-steve(the seat's linked user asGET /useranswers it; the card's assignee)Branch:
claude/project-thread-6uuq7f(the card thread's harness push branch, self-reported at 2026-10-27T16:41Z — the measured platform exception toclaude/issue-N-slug)Worktree:
objectstack-issue-26760Domain:
domain:skillsSeat:
domain:skills#1File surface:
skills/objectstack-formula/evals/cel-predicates-formulas.json(a positivelabel(record.kind)case and a negative predicate case withmust_not_contain: ['label(']), its row inscripts/check-skills-token-ratchet.mjs(trim duplicated wording first, then raise at the measurement under the standing ruling 「A:先削重复,余额按实测提高」), and a patch changeset for@objectstack/skills; nothing else (stop on breach; explain in the report)Container & model:
XS,mode:thread,model: CONTRACT_REVIEW_TIER(dispatch-gates --tier: MANDATORY —skills/**, clause ①, no one-line exit),effort: high. Tier H (skills/**): draft PR; ⛔ the dev never readies, arms or enqueues it.Clause-②: no
Metadata-protocol: none
Responsibility:
n/a — not a defect card(a follow-up owed by landed spec work)Thread-read: 6445256395
Premise: holds (seat re-measure on
origin/main8b276385: PR #26755 merged;skills/objectstack-formula/SKILL.md:196teacheslabel(record.f); no eval nameslabel(; the eval file sits at 1046/1046)Serial constraints cleared: the Blocked-by PR #26755 merged; no open PR names
skills/objectstack-formula/**objectstack-fleet commentedon Oct 27, 2026
os-dev-report
{ "issue": 26760, "status": "done", "branch": "claude/project-thread-6uuq7f", "pr": "https://github.com/objectstack-ai/objectstack/pull/26810", "head": "827f9351", "session": "session_01ADHkmPm7NR24Xypi61eyg8", "premise_still_valid": true, "summary": "Two eval cases join skills/objectstack-formula/evals/cel-predicates-formulas.json: case 6 asks for a contract_version record name built from its kind select and expects label(record.kind) in an F formula (must_contain label(record.kind) + F backtick; must_not_contain the literal == 'clean' ? ternary and {record.); case 7 asks for a validation rule and a visibleWhen on the same object and refuses label( in either (must_contain record.kind, visibleWhen, P backtick), expecting predicates over the stored value. Paid down first: four sentences in cases 1, 2, 3 and 5 that restated their own must_not_contain assertions (212 bytes = 53 units, 1046 -> 993); the two cases add 1438 bytes = 359 units; the ratchet row rises AT the measurement, 1046 -> 1352, headroom 0, arithmetic in the row comment and the PR body under 「A:先削重复,余额按实测提高」. Patch changeset for @objectstack/skills. Draft PR, Tier H (skills/**): not readied, armed or enqueued.", "line_budget": "whole file 55 -> 75 lines, 1046 -> 1352 tokens (4181 -> 5407 B); whole package sum of skills/*/SKILL.md 4606 -> 4606 lines, 56132 -> 56132 tokens (no SKILL.md moves); ratchet row 1046 -> 1352 pinned at the measurement", "files_changed": ["skills/objectstack-formula/evals/cel-predicates-formulas.json", "scripts/check-skills-token-ratchet.mjs", ".changeset/26760-formula-label-eval.md"], "tests": "No runtime code moves; no test is owed. Evidence the cases pin what origin/main 8b276385 does, not invented: (1) pnpm --filter @objectstack/formula exec vitest run --maxWorkers=2 src/option-label.test.ts -> exit 0, Test Files 1 passed (1), Tests 8 passed (8) (pins: refuses at run time when the context carries no option labels; compiles the one argument shape and refuses every other; '' for an empty select). (2) a probe against the built packages/formula/dist: the three eval expressions compile ok; the name formula evaluates to 'v2 · 清稿' with optionLabels and 'v2 · ' for a null kind; label(record.kind) == 'Clean' evaluated with no optionLabels (a predicate's context) answers ok:false, kind runtime, 'label(record.kind) cannot be answered: this evaluation context carries no option labels'. (3) JSON.parse over the file at head: 7 cases; every must_contain absent from its case's must_not_contain and present in its expected_output; cases 6 and 7 carry no must_not_contain string in their expected_output.", "gates": "At head 827f9351: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack derived 60 commands (--tier: MANDATORY, skills/** clause ①, no one-line exit; 53 changed lines vs 3000); each run with stdin closed, exit captured before any pipe: 60/60 exit 0, including node scripts/check-skills-token-ratchet.mjs ('cel-predicates-formulas.json is 1352 tokens (ceiling 1352; headroom 0)'; '81 authored bundle file(s) within their ceilings') and pnpm check:pm-dispatch-gates ('dispatch-gates self-test (fast tier): 1935 cases pass', 311.7s). Reconciliation: node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack --ran ran.txt -> exit 0, '60 derived, 60 run, 0 NOT-MEASURED, 0 UNRUN' (a DERIVED zero). Build for the dist-reading gates: os-verify-lock VERDICT command-exit 0 (turbo build ./packages/**, 75/75 tasks, 7m11s). CI families on the PR: not read; the seat reads the PR check-runs.", "dogfood_battery": "n/a — no `!` in the PR title and no (narrowing) arm", "mcp_calls": "3 — mcp__github__issue_read get (#26760), issue_read get_comments (#26760) x2; no MCP write", "api_writes": "1 — pr_create via scripts/pm/with-fleet.sh --via dispatch (fleet-write run 40547457829, PR #26810, body 7131 B read back identical, assignee os-steve); plus this os-dev-report comment via post-stamped.mjs (2); git push is not a REST write", "deviations": ["The worktree ../objectstack-issue-26760 is kept (node_modules not removed) until the PR lands, so a FAIL round can re-run gates without a cold rebuild; this container runs no parallel agent.", "The harness pushes the card thread's branch claude/project-thread-6uuq7f (the measured platform exception to claude/issue-N-slug), as the Claim names."], "open_questions": [ { "question": "Case 6 pins the hard-coded ternary as the literal string == 'clean' ? (the stored value the prompt names). A ternary over a differently spelled value would pass must_not_contain; the must_contain of label(record.kind) carries the teaching.", "options": ["A keep the one literal (compact, as filed)", "B add a second spelling such as ? 'Clean' (+ ~5 units, another ceiling move)"], "recommendation": "A — the positive assertion is the gate; the negative is a tripwire for the exact shape the card describes." } ], "out_of_scope_findings": ["Existing case 3's expected_output carries the string ISCHANGED( inside its own refusal sentence ('No ISCHANGED() function.' was one of the four cut sentences, so this is now resolved as a side effect of the paid-down cut; noted for the record, nothing left to file)."] }objectstack-fleet commentedon Oct 27, 2026
ACCEPT → PR #26810 comment 6448436708 ·
domain:skills#1· 2026-10-27T17:10Z · Tier H; draft until a GOVERNED_APPROVERS approval; contract review requested from the director seat.objectstack-fleet commentedon Oct 27, 2026
Landed ·
domain:skills#1·session_01ADHkmSrRBdu9MdC6fc33Uu· 2026-10-27T23:20Z — PR #26810 merged through the queue (merge commit40264beec4) on theos-zhuangapproval, after the director's PASS 6448530347;cel-predicates-formulas.jsonnow carries thelabel(record.kind)positive case and the predicate negative case, its ratchet row at the measurement. The card closed by itsFixesline;pm:dispatchedstripped in this act. The director's should-fix on case 4 is #26814, dispatched next.