fix(agent): stabilize post-merge release paths

skills/scripts/e2e/wizard-onebot-agent-runtime.mjs

@@ -81,7 +81,8 @@ async function api(page, path, options = {}) {
         headers: {
           Authorization: `Bearer ${token}`,
           "Content-Type": "application/json",
-          "X-Workspace-Id": localStorage.getItem("langbot_active_workspace_uuid") || "",
+          "X-Workspace-Id":
+            localStorage.getItem("langbot_active_workspace_uuid") || "",
         },
         body:
           options.body === undefined ? undefined : JSON.stringify(options.body),
@@ -205,9 +206,7 @@ try {
   result.visible_signals.push("bot-created", "adapter-enabled");
 
   await page.getByRole("button", { name: /Next|下一步|次へ/ }).click();
-  const localAgentTitle = page
-    .getByText(/^(Local Agent|本地 Agent)$/)
-    .first();
+  const localAgentTitle = page.getByText(/^(Local Agent|本地 Agent)$/).first();
   const localAgentCard = localAgentTitle.locator(
     'xpath=ancestor::*[@data-slot="card"][1]',
   );
@@ -316,6 +315,24 @@ try {
 } catch (error) {
   if (!["blocked", "env_issue"].includes(result.status)) result.status = "fail";
   result.reason = result.reason || error.message;
+  if (browser?.page && token && botId) {
+    const [routeStatus, botLogs] = await Promise.all([
+      api(
+        browser.page,
+        `/api/v1/platform/bots/${encodeURIComponent(botId)}/event-routes/status`,
+      ).catch(() => ({ status: 0, json: {} })),
+      api(
+        browser.page,
+        `/api/v1/platform/bots/${encodeURIComponent(botId)}/logs`,
+        {
+          method: "POST",
+          body: { from_index: -1, max_count: 50 },
+        },
+      ).catch(() => ({ status: 0, json: {} })),
+    ]);
+    result.runtime.failure_route_status = routeStatus.json.data || null;
+    result.runtime.failure_bot_logs = botLogs.json.data || null;
+  }
   if (browser?.page) await safeScreenshot(browser.page, paths.screenshot);
 } finally {
   if (browser?.page && token) {

skills/skills.index.json

@@ -1003,8 +1003,12 @@
             "rag"
           ],
           "automation": "scripts/e2e/langrag-kb-retrieve.mjs",
-          "setup_automation": [],
-          "setup_provides_env": [],
+          "setup_automation": [
+            "node:scripts/e2e/ensure-langrag-sentinel-kb.mjs --write-env"
+          ],
+          "setup_provides_env": [
+            "LANGBOT_LOCAL_AGENT_RAG_KB_UUID"
+          ],
           "evidence_required": [
             "ui",
             "screenshot",

skills/skills/langbot-testing/cases/langrag-kb-retrieve.yaml

@@ -25,6 +25,10 @@ automation_env:
 automation_env_any:
   - LANGBOT_LOCAL_AGENT_RAG_KB_UUID|LANGBOT_RAG_KB_UUID
 automation_expected_text: "azalea-cobalt-7421"
+setup_automation:
+  - "node:scripts/e2e/ensure-langrag-sentinel-kb.mjs --write-env"
+setup_provides_env:
+  - LANGBOT_LOCAL_AGENT_RAG_KB_UUID
 preconditions:
   - "LangRAG is installed and initialized in the active LangBot instance."
   - "A working embedding model is available, preferably chroma-all-MiniLM-L6-v2 for local repeatability."

skills/skills/langbot-testing/fixtures/complex-agent-task/verify.py

@@ -8,13 +8,20 @@ from pathlib import Path
 
 def has_initial_failure_section(report: str) -> bool:
     folded = report.casefold()
-    return any(
+    if any(
         marker in folded
         for marker in (
             "initial fail",
             "initially fail",
             "baseline fail",
         )
+    ):
+        return True
+
+    baseline_markers = ("baseline test", "before any edit", "before editing")
+    failure_markers = ("failing test", "failed test", "failures=", "errors=")
+    return any(marker in folded for marker in baseline_markers) and any(
+        marker in folded for marker in failure_markers
     )
 
 

skills/test/test_complex_agent_task_verify.py

@@ -26,9 +26,23 @@ class ComplexAgentTaskVerifyTests(unittest.TestCase):
     def test_accepts_baseline_failure_heading(self) -> None:
         self.assertTrue(MODULE.has_initial_failure_section("## Baseline failing tests\n- test_price"))
 
+    def test_accepts_baseline_run_heading_with_failures_in_body(self) -> None:
+        report = """\
+## Baseline test run (before any edits)
+
+The suite reported FAILED (failures=2). Failing tests:
+
+- test_price
+- test_inventory
+"""
+        self.assertTrue(MODULE.has_initial_failure_section(report))
+
     def test_rejects_report_without_failure_section(self) -> None:
         self.assertFalse(MODULE.has_initial_failure_section("## Verification\nAll tests pass."))
 
+    def test_rejects_passing_baseline_without_failure_evidence(self) -> None:
+        self.assertFalse(MODULE.has_initial_failure_section("## Baseline test run\nAll tests pass."))
+
 
 if __name__ == "__main__":
     unittest.main()