feat(lint): widen flow-trigger-unroutable to the omission shape (#7215)

[os-project-manager]

Fixes #7215

What changed

flow-trigger-unroutable (#6637) previously fired only on the contradiction shape:
config.triggerType present but the engine routes it to no trigger (triggerType: 'onCreate').

Per #7215's maintainer-delegated ruling (disposition 1, "take it now" — see the RULED comment on
that issue), this PR widens the same rule to also fire on the omission shape: type: 'record_change' with no triggerType key at all. Both shapes fall through
AutomationEngine.resolveTriggerBinding's same branch chain to undefined, and both are equally
invisible to getTriggerBindingAudit (which skips a flow with no resolved binding as "manual —
nothing to bind"). They are two spellings of one defect, so they share the same rule id and the
same error severity rather than getting a new id.

The rule's other exclusion is unchanged: a record_change flow that declares something the engine
does route (config.schedule, triggerType: 'api', a config.timeRelative object) still binds
and fires — on the wrong trigger's terms — and is a different defect this rule does not speak to.

Corpus count (measured before landing, per the dispatch's stop condition)

Ran the widened criterion over every type: 'record_change' flow definition found by content
search across examples/, apps/, and packages/ (excluding test fixtures and the CLI's
scaffold-string template, which produces a different, non-record_change shape). The tree has
exactly two real record_change flows:

• examples/app-todo/src/flows/task.flow.ts — TaskCompletionFlow, repaired by fix(example-todo): arm task_completion — it bound to nothing and gated on a key nothing reads (#6882) #7039 to declare
  triggerType: 'record-after-update'.
• examples/app-showcase/src/automation/flows/index.ts — UrgentTaskAlertFlow, declares
  triggerType: 'record-after-write'.

Both already declare an explicit, valid triggerType. Verified programmatically (not just by
grep) by importing both flow modules and running validateFlowTriggerReadiness against them
directly:

TOTAL_OMISSION_INSTANCES=0

Omission instances found: 0. The widening ships green on the whole tree — no baseline entry,
no example-app churn, no new vocabulary.

Test changes

validate-flow-trigger-readiness.test.ts contained a scope-boundary pin
(an ABSENT triggerType — dead too, deliberately deferred (#6637 corpus)) whose own comment said
widening "is a deliberate edit that has to delete this test, not a side effect." That test is
deleted, per #7215, and replaced with:

• a test that the omission shape now fires (FLOW_TRIGGER_UNROUTABLE, error, same path shape as
  the contradiction case),
• a test that the rule id/severity/path are identical whether the token is wrong or missing,
• a test that an omission with a routed sibling key (config.schedule / config.timeRelative)
  still stays silent — the "mis-bound, not un-bound" exclusion applies to the omission shape too,
• a test that non-record_change flow types (autolaunched, screen, schedule, api, no
  type at all) stay silent even with triggerType absent — enumerated from the code's exact
  flow.type === 'record_change' guard, not guessed.

All of the contradiction shape's pre-existing tests are untouched and still pass.

Provenance comments were updated in both files (the rule's own scope-boundary comment above the
1f block, and the exported FLOW_TRIGGER_UNROUTABLE JSDoc) to record the widening's history —
kept in historical form per the PR #7172 discipline, with the forward-pointer to "the still-open
widening question" now resolved to #7215 rather than deleted outright.

Reverse verification

Temporarily restored the pre-widening rule file (git show origin/main:… >, no git stash) and
re-ran the new tests against it: all 4 new omission-shape tests went red, while the other 61
pre-existing tests (contradiction shape, severity map, etc.) stayed green — confirming the new
tests are non-vacuous and the existing scope is untouched. Restored the widened file and re-ran;
back to 65/65 green.

Verification

pnpm --filter '@objectstack/lint^...' build          # deps (spec, formula) — OK
cd packages/lint && npx vitest run                    # 69 files, 1808 passed
cd packages/lint && npx tsc --noEmit                   # clean
cd packages/cli && npx vitest run test/authoring-rule-command-parity.test.ts test/doctor-refs.test.ts
                                                        # 2 files, 19 passed
node scripts/check-nul-bytes.mjs                       # OK

Changeset

Added .changeset/lint-flow-trigger-unroutable-omission.md (@objectstack/lint: minor) — a
functional widening of a published lint rule's gating criterion, precedent PR #7214.

---

Generated by Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
objectstack	Ignored		Aug 10, 2026 4:21am

[github-actions]

📓 Docs Drift Check

This PR changes 1 package(s): @objectstack/lint.

2 hand-written doc(s) reference the affected code and may need an implementation-accuracy re-verification:

• content/docs/automation/hook-bodies.mdx (via @objectstack/lint)
• content/docs/permissions/authorization.mdx (via @objectstack/lint)

⛔ 1 release-owned page(s) also reference the affected code. These are read-only:

• content/docs/releases/v17.mdx (via @objectstack/lint)

content/docs/releases/ is RELEASE-OWNED (AGENTS.md "Documentation Guardrails"): release
notes are written centrally at release time, and a code PR that edits them is the exact PR
that guardrail exists to stop. They are still audited — read-only. If one of them is actually
wrong, file an issue or open a dedicated docs-only PR; do not edit it here.

Advisory only. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs origin/main → pass the list as args.docs.

[os-project-manager]

PM 验收:ACCEPT。 门禁按名点名复读(head 1bb3cd8fa):25 项全部 success/skipped,零 failure(ESLint 04:25:55Z、TypeScript Type Check 04:37:03Z)。已 draft:false + auto-merge 入队。

⚠️ 验收方式说明(据实记录):本单的 dev agent 在工作进程重启中丢失,结构化报告从未送达。承重项因此由 PM 直接对 diff 复核,而非采信报告 —— 下列每条都是我自己在 origin/main...1bb3cd8fa 上读出来的:

• corpus 计数在 PR 正文且为 0(非零本应停手):全库仅两个真实 record_change flow —— examples/app-todo 的 TaskCompletionFlow(fix(example-todo): arm task_completion — it bound to nothing and gated on a key nothing reads (#6882) #7039 已修为 record-after-update)与 examples/app-showcase 的 UrgentTaskAlertFlow(record-after-write),两者均已显式声明。且系导入两个 flow 模块跑生产入口 validateFlowTriggerReadiness 所得,非 grep。
• 判据改动面最小:if 条件只删掉 config.triggerType != null 一个合取项;routesToSomeTrigger 排除链字符未动 —— 「绑错」而非「没绑上」的第二种形态仍然不判,这是本次收紧最容易越界的地方,没越。
• 消息按 hasTriggerType 分叉,不会对 undefined 调 renderTriggerToken。
• pin 测试蓄意删除且引用 [decision] Should flow-trigger-unroutable widen to the absent-triggerType (omission) shape — #7041 item 2, decide together with #6041 #7215:原测试自述「收紧是必须删掉这个测试的有意行为,不是顺手带过的副作用」,该行已随测试移除,原位替换为引用 [decision] Should flow-trigger-unroutable widen to the absent-triggerType (omission) shape — #7041 item 2, decide together with #6041 #7215 的遗漏形测试节。
• 反向验证 —— 本单最硬的一条:还原收紧前的规则文件后重跑,4 条新测试全红、61 条既有测试全绿,证明新测试非空转且既有判据面未被扰动。且还原用的是 git show origin/main:… > 而非 git stash —— 守住了共享 stash 栈的红线。
• 前向指针按 fix(lint): correct stale provenance comment in validate-flow-trigger-readiness.test.ts #7172 纪律更新:规则 JSDoc 与 1f 注释块保留历史形态,把「仍未决的收紧问题」指向 [decision] Should flow-trigger-unroutable widen to the absent-triggerType (omission) shape — #7041 item 2, decide together with #6041 #7215 已裁决,而非直接删除。
• 真 changeset(@objectstack/lint: minor)。
• 清理:worktree 洁净移除,remove 未带 --force。

合入后本单收卡、#7215 清 pm:dispatched。

---

Generated by Claude Code