fix: list-search seed and authoring e2e drop their clipboard tail (Linux CI)

PR #11's three red shards traced to ONE cause (confirmed via CI
artifacts): the new seed and the configure-lists authoring test both
ended in apply-clipboard-write-text, which errors on Linux (no
pbcopy/osascript) — the exact environment class the earlier triage
established. Fixed by REMOVING the unneeded node (neither test needs
clipboard I/O to prove list-search; the seed now ends at list-search
itself, the same warn-only Process-leaf shape list-lookup's sibling
seed already uses). Shard 2 was a fail-fast cancellation casualty
(50/50 green — matrix fail-fast:false rides the next PR); the codeexec
blip was the documented pre-existing flake class, evidenced, untouched.

Takeover note: the delegated agent stopped silently mid-verification
(third such instance tonight); fix verified and landed by the
orchestrator directly. The recurring 'linker contention' commit-gate
failures root-caused to DISK FULL, not contention — corrected in the
session memory.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FYwojT8GdUbYSoggbvEFft

Author: alicoding

frontend/e2e/configure-lists.spec.ts

@@ -109,12 +109,18 @@ test('list-search node: configuring a real match parameter through the Inspector
   const panel = activePanel(page)
   await panel.getByLabel('Label').fill('E2E list-search config test')
 
+  // Deliberately no apply-clipboard-write-text terminal node -- this
+  // test proves the list-search Inspector's own authoring/execution,
+  // not clipboard I/O, and a clipboard apply step has no clipboard on
+  // a headless Linux CI runner (docs/SPEC.md §1.3; the same fix
+  // applied to the seeded "Example: Country lookup (search)" workflow
+  // after a real CI failure, builtinworkflows_list.go). Ending at
+  // list-search itself is an accepted, warn-only Process leaf
+  // (ADR-0028).
   await panel.getByTestId('toggle-palette').click()
   await dragPaletteItemToCanvas(page, 'list-search')
-  await dragPaletteItemToCanvas(page, 'apply-clipboard-write-text')
-  await expect(panel.locator('.react-flow__node')).toHaveCount(3)
+  await expect(panel.locator('.react-flow__node')).toHaveCount(2)
   await connectNodes(page, 'Trigger: manual', 'List: search')
-  await connectNodes(page, 'List: search', 'Apply: write plain text to clipboard')
 
   await clickCanvasNode(page, panel, 'List: search')
   const inspector = panel.getByTestId('composition-inspector')

internal/domain/composition/builtinworkflows_list.go

@@ -45,12 +45,19 @@ func builtInListWorkflows() []Workflow {
 	// Demonstrates a typed Object result (results/matched/first_match/
 	// match_count/list_id) a downstream step could branch on, not just
 	// a single scalar Attribute -- list-lookup's own seed above stays
-	// untouched, proving the two coexist.
+	// untouched, proving the two coexist. Deliberately ends AT
+	// list-search itself, mirroring list-lookup's own seed above,
+	// rather than adding a terminal apply-clipboard-write-text step:
+	// that step has no clipboard on a headless Linux CI runner
+	// (docs/SPEC.md §1.3) and would only be exercising clipboard I/O
+	// this seed isn't actually about -- caught by a real CI failure
+	// (goal 0011's own PR), not assumed. A Process leaf is an accepted,
+	// warn-only ending (ADR-0028), the same shape several other seeds
+	// already use.
 	const (
 		listSearchTriggerID = "example-list-search-trigger"
 		listSearchCaptureID = "example-list-search-capture"
 		listSearchStepID    = "example-list-search-step"
-		listSearchApplyID   = "example-list-search-apply"
 	)
 	const listSearchMatchParams = `[{"column":"code","value":"attr:code","matchType":"exact"}]`
 	listSearchNodes, err := ResolveNodeDefaults([]Node{
@@ -63,7 +70,6 @@ func builtInListWorkflows() []Workflow {
 				"matchParams":     listSearchMatchParams,
 				"outputAttribute": "searchResult",
 			}},
-		{ID: listSearchApplyID, NodeTypeID: "apply-clipboard-write-text", Position: Position{X: 0, Y: 300}},
 	})
 	if err != nil {
 		panic("built-in workflow references an unknown node type: " + err.Error())
@@ -88,7 +94,7 @@ func builtInListWorkflows() []Workflow {
 		{
 			ID:          "example-list-search-workflow",
 			Label:       "Example: Country lookup (search)",
-			Description: "Captures a typed 'code' Attribute and searches the seeded \"Example: Country codes\" List (Configure > Lists) via list-search -- an exact match on its 'code' column, writing a typed Object result ({results, matched, first_match, match_count, list_id}) into 'searchResult', then writes the captured code to the clipboard. Unlike list-lookup's plain scalar output, this demonstrates the richer typed result a downstream step (e.g. a Branch condition on searchResult.matched) could reference. Run it with code = US, CA, MX, or FR to see a match -- SU is a deliberately Expired seed row, excluded from matching by default (docs/goals/0011-lists-maturation.md).",
+			Description: "Captures a typed 'code' Attribute and searches the seeded \"Example: Country codes\" List (Configure > Lists) via list-search -- an exact match on its 'code' column, writing a typed Object result ({results, matched, first_match, match_count, list_id}) into 'searchResult'. Unlike list-lookup's plain scalar output, this demonstrates the richer typed result a downstream step (e.g. a Branch condition on searchResult.matched) could reference. Run it with code = US, CA, MX, or FR to see a match -- SU is a deliberately Expired seed row, excluded from matching by default (docs/goals/0011-lists-maturation.md).",
 			Nodes:       listSearchNodes,
 			Attributes: []AttributeDef{
 				{Key: "code", Label: "Code", Type: FieldText},
@@ -97,7 +103,6 @@ func builtInListWorkflows() []Workflow {
 			Edges: []Edge{
 				{ID: "example-list-search-e0", Source: listSearchTriggerID, Target: listSearchCaptureID},
 				{ID: "example-list-search-e1", Source: listSearchCaptureID, Target: listSearchStepID},
-				{ID: "example-list-search-e2", Source: listSearchStepID, Target: listSearchApplyID},
 			},
 			BuiltIn: true,
 		},

internal/services/executionsvc/listsearch_seed_test.go

@@ -61,11 +61,14 @@ func TestSeededListSearchExample_Match_WritesTypedResult(t *testing.T) {
 	if summary.Status != "SUCCESS" {
 		t.Fatalf("RunWorkflow(code=US) status = %q, want SUCCESS -- error: %s", summary.Status, summary.Error)
 	}
-	// The seed's final step (apply-clipboard-write-text) writes the
-	// captured code itself -- list-search's own typed result lives in
-	// the 'searchResult' Attribute, not the string Payload/Output,
-	// proven at the composition-unit-test layer
-	// (listsearch_test.go) rather than re-asserted here.
+	// The seed ends AT list-search itself (no terminal apply step --
+	// a real Linux-CI clipboard failure caught during goal 0011's own
+	// PR is why, see builtinworkflows_list.go), so the workflow's
+	// final Payload/Output is still whatever capture-attribute set it
+	// to; list-search's own typed result lives in the 'searchResult'
+	// Attribute, not the string Payload/Output, proven at the
+	// composition-unit-test layer (listsearch_test.go) rather than
+	// re-asserted here.
 	if summary.Output != "US" {
 		t.Errorf("RunWorkflow(code=US) output = %q, want %q", summary.Output, "US")
 	}
@@ -77,8 +80,7 @@ func TestSeededListSearchExample_NoMatch_WritesUnmatchedResult(t *testing.T) {
 	// Unlike list-lookup's onMiss="fail" default, list-search never
 	// fails the run on a miss -- it always writes a typed
 	// {matched:false, results:[], ...} object and lets the workflow
-	// continue (a Decision downstream would branch on it). The seed's
-	// own apply-clipboard-write-text step still runs.
+	// continue (a Decision downstream would branch on it).
 	summary, err := exec.RunWorkflow(wfID, RunKindTest, map[string]string{"code": "ZZ"})
 	if err != nil {
 		t.Fatalf("RunWorkflow: %v", err)