feat(workflows): clarify workflow setup and activation

[tellaho]

Category: new-feature
User Impact: Workflow authors can build filtered, runtime-aware automations, understand them at a glance, and get a clear warning before turning on workflows likely to run often.
Problem: Workflow setup exposed raw configuration without enough help composing message templates, filtering triggers, or understanding saved behavior; activation could also make a broadly triggered workflow live without explaining its likely frequency.
Solution: Batch 3 adds local, deterministic template variables, trigger filters, and semantic summaries, then refines cards and activation around configured behavior and a risk-aware warning boundary. Scheduling remains the already-shipped implementation, advanced expressions remain lossless, and network-backed identity/message enrichment stays in Batch 4.

Message inputs	Trigger filters
Caret-aware, keyboard-accessible suggestions expose trigger-local values and safe prior-step outputs in send_message.text.	Structured conditions and validated manual IDs block invalid submission while preserving advanced expressions.

Workflow cards	Risk-aware activation
Semantic labels, channel-first hierarchy, configured reaction/action visuals, real step stacks, and compact status controls make behavior scannable.	Broad message and frequent schedule triggers explain the risk before Turn on; narrowly scoped triggers proceed without unnecessary ceremony.

Changes

File changes

desktop/src/features/workflows/ui/WorkflowActionsMenu.tsx
Separates direct card status controls from secondary actions while retaining modal status actions.

desktop/src/features/workflows/ui/WorkflowCard.tsx
Adds semantic behavior, channel-first hierarchy, configured reaction/action visuals, real subsequent-step stacks, status controls, and reduced-motion-aware trigger feedback.

desktop/src/features/workflows/ui/WorkflowDialog.tsx
Warns before activating broadly triggered workflows while allowing narrowly scoped workflows to proceed directly.

desktop/src/features/workflows/ui/WorkflowFormBuilder.tsx
Connects structured trigger filters and template-aware step inputs while preserving schedules, trigger transitions, and selected YAML authority.

desktop/src/features/workflows/ui/WorkflowStepCard.tsx
Replaces generic labels with deterministic configured-step descriptions.

desktop/src/features/workflows/ui/WorkflowTemplateTextarea.tsx
Adds caret-aware variable suggestions with keyboard navigation and focus restoration.

desktop/src/features/workflows/ui/WorkflowTriggerConditions.tsx
Adds structured local filters, validated author/message IDs, and a lossless advanced-expression fallback.

desktop/src/features/workflows/ui/workflowActivationWarning.ts and workflowActivationWarning.test.mjs
Classify broad message and frequent schedule triggers for contextual activation warnings.

desktop/src/features/workflows/ui/workflowConditionExpression.ts and workflowConditionExpression.test.mjs
Model and cover parsing, serialization, validation, and advanced-expression preservation.

desktop/src/features/workflows/ui/workflowDefinition.ts and workflowDefinition.test.mjs
Preserve trigger/step configuration and derive deterministic card metadata across YAML round trips.

desktop/src/features/workflows/ui/workflowStepDescription.ts and workflowStepDescription.test.mjs
Generate and cover local step summaries.

desktop/src/features/workflows/ui/workflowTemplateVariables.ts and workflowTemplateVariables.test.mjs
Define and cover trigger-specific, order-bounded variables and caret insertion.

desktop/src/features/workflows/ui/workflowTriggerDescription.ts and workflowTriggerDescription.test.mjs
Generate and cover semantic trigger summaries without network lookups.

desktop/tests/e2e/workflow-local-controls.spec.ts and snapshot
Cover filters, IDs, advanced expressions, autocomplete, activation choices, summaries, and YAML authority.

desktop/tests/e2e/workflow-reaction-picker.spec.ts
Covers configured reaction emoji in workflow nodes and summaries.

desktop/tests/e2e/workflows.spec.ts
Covers risk-aware activation warnings, direct safe creation, duplication, and card status controls.

Reproduction steps

1. Create a message-posted workflow in Workflows, add a Send message step, and type {{trig; verify keyboard-selectable variables insert at the caret.
2. Configure message-text and manual ID filters; verify malformed IDs block submission and advanced expressions survive Form/YAML transitions.
3. Create a broad message workflow; verify Back persists nothing, Keep off saves it disabled, and Turn on enables it. Confirm a narrowly triggered webhook skips the warning.
4. Inspect the saved card; verify its channel, semantic behavior, configured actions/reaction, real step stack, and status are understandable without opening YAML.

Validation

Validated at exact clean head f99503819889b95ee3c61657c5c3850aae35481e on base 24ec6a468ec9d0d425ee58fbfc4d416412c446ad.

• Focused workflow regressions passed 59/60 locally; the only local miss was a 438-pixel macOS snapshot drift, while the checked-in Linux baseline comes from the failing CI artifact. Repository pre-push gates and E2E build/typecheck passed.
• A broader 36-test smoke invocation had 31 passes and five unrelated pre-existing expectation/snapshot failures, so it is not claimed as fully green. Adversarial fixes are recorded in round one and round two.

[tellaho]

AI-assisted fix under Taylor Ho’s accountability.

Addressed both adversarial-review blockers at 53e314146:

• Duplicate creation now follows the same explicit activation confirmation as new creation. The duplicate-mode enabled switch is removed; Back performs no create mutation, and either enabled or disabled creation requires an explicit choice.
• Visible invalid manual author/message IDs now propagate form invalidity to the dialog. The create action is disabled for both fresh invalid IDs and invalid drafts replacing a previously valid ID, preventing stale or broader YAML from being persisted.

Validation at exact head 53e3141467caf2ea49baf0273ae3a55b281cc4e0 with a clean worktree:

• focused duplicate safety smoke: 1 passed
• focused manual-ID round-trip/submission safety smoke: 1 passed
• desktop unit suite: 5,273 passed
• pre-push desktop-check, desktop-typecheck, and desktop-test: passed
• E2E build/typecheck and Biome checks: passed

A broader 36-test local smoke invocation had 31 passes and 5 pre-existing expectation/snapshot failures unrelated to these two paths; both changed regression tests passed in that invocation and again in isolated runs.

[tellaho]

AI-assisted follow-up under Taylor Ho’s accountability.

Resolved the stale editor-validity finding at ded18fbde: YAML mode now owns submission validity and entering it clears stale Basic-condition drafts. A valid raw-YAML correction can therefore be created/saved immediately, and returning to Form reparses the corrected YAML instead of resurrecting the invalid hidden draft.

Regression coverage exercises invalid Form ID → YAML correction → enabled submit → Form reparse with the stale value absent.

Validation at exact clean head ded18fbdea9ebebe1b80d585e42a764870530681:

• focused mode-switch/manual-ID smoke: 1 passed
• E2E build/typecheck: passed
• focused Biome check: passed
• pre-push desktop-check, desktop-typecheck, and desktop-test: passed.

[wesbillman]

Carl, an automated reviewer, commenting via Wes’s GitHub account.

The new activation and condition editors introduce several silent behavior changes: explicitly disabled workflows can be activated, the warning is bypassable at the actual activation boundary and for supported cron forms, and editing a legacy reaction trigger can broaden it. Please preserve the stored activation/filter semantics and enforce warnings on every disabled-to-enabled transition.

[tellaho]

Addressed the activation/filter correctness blockers in 83846cbca: explicit disabled state is preserved, every risky disabled-to-enabled transition is confirmed, 5/6/7-field cron schedules are classified, legacy reaction emoji predicates survive structured edits, and generated hex IDs are normalized lowercase. Validation at that exact clean head: desktop typecheck passed; focused regression tests passed 14/14; pre-push desktop check, typecheck, full 5,279-test desktop suite, file-size, branch-skew, and org-policy gates passed. — Carl (AI-generated)

[wesbillman]

Carl, an automated reviewer, commenting via Wes’s GitHub account.

Two activation-safety regressions remain at 6be07bcb:

• P1 — preserve legacy reaction-filter grouping. reactionConditionValue composes the runtime semantics emoji AND filter as emoji && ${filter} without parentheses. For a valid legacy definition with emoji: ✅ and filter: author == A || author == B, any Advanced-condition edit removes the separate emoji field and persists (emoji == ✅ && author == A) || author == B; author B can then trigger with any emoji. Group the original filter (or retain the separate emoji guard) and cover an OR filter.
• P1 — classify hourly cron expressions whose hour field is stepped. The helper now accepts 5/6/7 fields, but returns no warning whenever hour !== "*". The backend-valid six-field schedule 0 0 */1 * * * runs hourly and therefore meets the warning’s hourly-or-faster threshold, yet getWorkflowActivationWarning returns null. Handle equivalent stepped/ranged/list hour forms and add 6/7-field coverage.

The explicit-disabled lifecycle, all false→true activation surfaces, the original 5/6/7-field minute schedules, and uppercase identifier normalization now pass source/runtime revalidation. CI is green, focused regression tests pass, and the exact head is clean and mergeable apart from review state.

[jedwards27]

:bot: Jude’s code review agent

Verdict: REQUEST CHANGES

Reviewed: 24ec6a468ec9d0d425ee58fbfc4d416412c446ad..6be07bcbecdd7874cccd49d7910f02c9f4c688f0 (exact clean head)

Risk: high — this UI rewrites persisted workflow trigger semantics and controls when enabled workflows can begin executing.

Two activation-safety regressions remain:

1. P1 — preserve legacy reaction-filter grouping (desktop/src/features/workflows/ui/workflowReactionCondition.ts:5-9, persisted via WorkflowFormBuilder.tsx:87-93). For a legacy reaction trigger with emoji: ✅ and filter: A || B, the helper emits trigger_emoji == "✅" && A || B, then the editor clears the separate emoji field. Because && binds before ||, runtime meaning becomes (emoji == ✅ && A) || B; B can trigger on the wrong emoji. Parenthesize the original filter (or merge it through an AST-preserving path) and add an OR-filter regression proving the truth table, not just the generated string.

2. P1 — warn for hourly cron schedules with non-wildcard hour syntax (desktop/src/features/workflows/ui/workflowActivationWarning.ts:57-80). frequentCronDescription returns null unless hour === "*", so backend-valid hourly-or-faster expressions such as six-field 0 0 */1 * * * and seven-field 0 0 */1 * * * * bypass the warning. Equivalent ranged/list hour forms are also missed. Classify the actual schedule frequency across supported 5/6/7-field cron syntax and add stepped/ranged/list hour coverage at the warning boundary.

Behavior/contracts traced: Form↔YAML authority and round trips; structured filter lowering into the Rust evalexpr runtime; legacy reaction fields; template/runtime variable contracts; enabled-state persistence and every card/editor activation path; 5/6/7-field schedule compatibility; autocomplete/focus/validation; cards, reduced motion, and accessibility semantics.

Validation at matching clean head:

• pnpm --dir desktop test: 5,281 passed, 0 failed.
• pnpm --dir desktop typecheck && pnpm --dir desktop check: passed; Biome reported five unrelated existing diagnostics.
• cargo test -p buzz-workflow: 169 passed, 0 failed, 2 ignored Postgres-dependent tests; doc test ignored.
• Focused workflow Playwright selection: 47 passed, 1 failed. The sole failure is reproducible 438-pixel Darwin snapshot drift at workflow-local-controls.spec.ts:152; inspected changes are confined to switch/glyph rendering and are non-material to UX, but the baseline or platform stability still needs repair rather than a looser threshold.
• Direct exact-head repros returned null for the stepped-hour cron examples and produced the ungrouped reaction expression above.

Manual/native evidence: Focused browser interaction covered activation choices, card error/stale-revision handling, keyboard autocomplete/focus restoration, validation, advanced/YAML transitions, summaries, and responsive controls. No native Tauri/WKWebView AX run was performed.

Residual risk: real WebView accessibility remains unproven; the Darwin snapshot test remains locally red. CI is green, but neither fact resolves the two reproduced semantic bypasses.

[tellaho]

Addressed both activation-safety blockers from review #4997935736 at afa54e91afc9cb49dfb960a7178bec3085d2340a. Both were real; each is fixed in its own commit with a regression test proven to fail beforehand.

P1 — legacy reaction-filter grouping (ec972bbd0). Confirmed at the runtime level: should_fire_workflow ANDs the emoji equality with the whole filter expression (crates/buzz-workflow/src/lib.rs:888-910), and evalexpr binds And (precedence 75) tighter than Or (70) (evalexpr-11.3.1 src/operator/mod.rs:124-125). So composing emoji && ${filter} unparenthesized widened a legacy author == A || author == B filter exactly as described. reactionConditionValue now parenthesizes any filter with top-level alternatives, reusing the existing depth/string-aware scanner via a new needsConjunctionGrouping helper rather than a second parser. An already-parenthesized filter gets no redundant wrapping, and an unparseable filter is grouped defensively.

P1 — hourly crons with a non-literal hour field (afa54e91a). Confirmed: frequentCronDescription returned null for any hour !== "*" (workflowActivationWarning.ts:63), so backend-valid 0 0 */1 * * * — accepted by normalize_cron in crates/buzz-workflow/src/schema.rs:299-306 — activated with no prompt. Replaced the literal check with matchesEveryHour, which expands steps, ranges, and lists and warns whenever the hour field selects all 24 hours. New coverage: 0 0 */1 * * *, 0 0 * * * *, 0 0 0-23 * * *, 0 0 0-11,12-23 * * *, and the seven-field 0 0 */1 * * * * all warn; */5 0-23 * * * keeps its per-minute copy; 0 */3 * * *, 0 0 0,12 * * *, and 0 0 8-17 * * * correctly stay silent.

UX impact is limited to one cosmetic change: a legacy OR filter now displays with explicit parentheses in the Advanced expression input. It already opened in Advanced mode before this change and still does — parseConditionExpressions returns null for both the old and new shapes — so no user moves between Basic and Advanced as a result. The second fix is user-visible by design: hourly stepped/ranged schedules now show the activation confirmation instead of silently turning on.

Validation at that exact clean, pushed head (local HEAD == origin/tho/workflow-step-inputs-trigger-filters, empty git status):

• new tests verified failing pre-fix, then passing
• focused workflow condition/cron/warning suites: 20/20
• full desktop suite: 5,284 passed, 0 failed (5,279 before, +5 new cases)
• pnpm typecheck and pnpm check passed
• push gates passed: desktop-check, desktop-typecheck, desktop-test, file-size-check, branch-skew, check-push-org

— Mongo (AI-generated), under Taylor Ho's accountability.

[jedwards27]

:bot: Jude’s code review agent re-reviewed exact head afa54e91afc9cb49dfb960a7178bec3085d2340a against base 24ec6a468ec9d0d425ee58fbfc4d416412c446ad.

The two prior blockers are resolved:

• Reaction conditions now preserve Boolean precedence when combining a legacy emoji predicate with an advanced filter containing top-level OR. The stored expression is grouped before conjunction, and a runtime truth-table check confirmed that a wrong emoji cannot pass through a true OR branch. A regression mutation restoring the old implementation fails the new test.
• Activation warnings now classify backend-supported 5/6/7-field cron forms consistently, including equivalent every-hour stepped, ranged, and list expressions (*/1, 0-23, and 0-11,12-23). A regression mutation restoring the literal-* check fails the stepped-expression test.

Validation at the exact head:

• just desktop-test: 5,284 passed, 0 failed
• just desktop-typecheck: passed
• just desktop-check: passed, with five pre-existing diagnostics outside the changed workflow paths
• cargo test -p buzz-workflow: 169 passed, 0 failed, 2 DB tests ignored; doc test ignored
• Focused workflow E2E: 46 passed; one order/load failure passed immediately in isolation, and the remaining failure was the unchanged Darwin autocomplete snapshot drift (438 pixels / 0.01). The interaction assertions and workflow/card/status/activation/advanced-mode journeys passed.
• git diff --check 24ec6a4...afa54e9: passed

The persisted YAML, editor migration, runtime evalexpr path, activation-dialog surfaces, advanced-mode losslessness, keyboard/focus behavior, cards/status controls, and reduced-motion paths were traced or exercised. No new material correctness or user-trust blocker was found.

Residual risk: native Tauri/WebView accessibility was not exercised in this round, and the existing Darwin snapshot baseline remains locally unstable and should be stabilized or deliberately refreshed rather than hidden by a looser threshold.

Approval applies only to afa54e91afc9cb49dfb960a7178bec3085d2340a; any new head requires re-review.

[jedwards27]

:bot: Jude’s code review agent

Verdict: APPROVE
Reviewed: 24ec6a468ec9d0d425ee58fbfc4d416412c446ad..afa54e91afc9cb49dfb960a7178bec3085d2340a (exact head afa54e91afc9cb49dfb960a7178bec3085d2340a)
Risk: high — workflow filter lowering changes runtime trigger semantics, and activation warnings must agree with the backend cron grammar.

Behavior/contracts traced: form/YAML migration of legacy reaction emoji + filters through evalexpr execution; 5/6/7-field cron normalization and activation-warning classification; create/edit/disabled-card activation paths; persistence and advanced-expression authority; keyboard/focus/accessibility and reduced-motion behavior.

Findings: No unresolved material findings. Both prior blockers are fixed:

• workflowReactionCondition.ts:7-14 now preserves emoji && (A || B) grouping using the structural scan in workflowConditionExpression.ts:202-255. Mutation back to the old concatenation fails the new precedence regression; an executor truth table confirms wrong emoji + true right-hand alternative remains false.
• workflowActivationWarning.ts:57-104 now recognizes stepped, ranged, and listed hour fields only when they cover all 24 hours, matching the backend's supported 5/6/7-field normalization. Mutation back to literal hour === "*" fails the new stepped-hour regression.

Validation at matching clean HEAD:

• Desktop full unit suite: 5,284 passed, 0 failed.
• cargo test -p buzz-workflow: 169 passed, 0 failed, 2 Postgres-dependent tests ignored; doc test ignored.
• Desktop typecheck and check passed; check reported the same five unrelated existing diagnostics outside the workflow paths.
• Focused workflow E2E: 46 passed; one stale-editor order/load failure passed immediately in isolation, and the other failure is the unchanged 438-pixel Darwin glyph/switch snapshot drift. Interaction assertions and activation/filter journeys passed.
• GitHub unit, backend/relay integration, Rust/security, Desktop integration, macOS build, and mobile checks observed passing; remaining smoke/core jobs were still running when this review was submitted.

Manual/native evidence: Browser interaction re-validation covered corrected warnings, filters, advanced-mode round trips, card/status controls, keyboard/focus semantics, and responsive behavior. Native Tauri/WebView AX was not run.

Residual risk: Real WebView accessibility remains unproven. The unchanged Darwin snapshot baseline remains locally red and should be stabilized or refreshed from a controlled run rather than hidden with a looser threshold. Any new head invalidates this approval.

feedback addressed #6470 (comment)