@browserstack/mcp-server 1.5.0-beta.16 → 1.5.0-beta.18

package/capability/loadtesting.capability-index.json

@@ -161,6 +161,8 @@
         "intent": "List the load tests in a project — use this to find a testId or browse tests, optionally filtered by type, framework, tag or recent activity.",
         "guidance": [
           "Resolve projectId via listLoadTestProjects first.",
+          "To find a test by name, pass search=<name> so the server matches it in one call — do not page the whole project and match names yourself. In a large project this is the difference between one call and many.",
+          "If several tests match the name, disambiguate by lastRun / updatedSince — do not guess. A descriptive or auto-generated name (e.g. Automation_lts_*) is a real test, not noise; never silently drop it in favour of one that merely sounds more relevant.",
           "The numeric testId returned here is what getLoadTest and startLoadTestRun need — not a runId.",
           "lastRun summarises the most recent execution when present."
         ],
@@ -189,6 +191,49 @@
         "paginated": true,
         "max_page_size": 100
       },
+      {
+        "name": "searchLoadTests",
+        "method": "GET",
+        "path": "/api/v1/agent/loadTests/search",
+        "mode": "read",
+        "entity": "loadTest",
+        "query": [
+          {
+            "name": "name",
+            "type": "string",
+            "description": "Test name to search for (substring match, case-insensitive); required."
+          },
+          {
+            "name": "limit",
+            "type": "integer",
+            "description": "Max items to return (1-100, default 50)."
+          }
+        ],
+        "intent": "Find a load test by name across ALL of the caller's projects — use this to resolve a named test to a testId without knowing which project it is in.",
+        "guidance": [
+          "Group-scoped: unlike listLoadTests it needs no projectId. Prefer it to resolve a named test in one call instead of paging every project.",
+          "Each row carries testId, projectId, name, testType and framework — disambiguate same-named tests in different projects by projectId.",
+          "Returns one row per test (its latest version); the numeric testId is what getLoadTest and startLoadTestRun need."
+        ],
+        "returns": [
+          "tests",
+          "hasMore"
+        ],
+        "responses": {
+          "200": {
+            "$response": "Success"
+          },
+          "400": {
+            "$response": "BadRequest"
+          },
+          "401": {
+            "$response": "Unauthorized"
+          },
+          "500": {
+            "$response": "InternalServerError"
+          }
+        }
+      },
       {
         "name": "getLoadTestProjectTrends",
         "method": "GET",
@@ -1156,6 +1201,8 @@
           "testId is numeric; runId values returned are UUIDs.",
           "Each run carries a tags array — its own run-level tags if set (via updateLoadTest with runId), otherwise the tags inherited from the test. Filter with the tag query param.",
           "Filter with status / verdict / date range; page with cursor + limit.",
+          "To find a run by date, pass sinceIso/untilIso to narrow server-side — do not list the full history and scan it yourself. To compare two dated runs, make one narrowed call per date rather than one broad listing.",
+          "If a date resolves to more than one run, ask the user which; do not assume the first or the latest.",
           "One call returns this test's run history — reuse it and page with cursor when needed; do not re-list the same window."
         ],
         "returns": [
@@ -1586,6 +1633,7 @@
         ],
         "capabilities": [
           "listLoadTests",
+          "searchLoadTests",
           "getLoadTest",
           "createLoadTest",
           "updateLoadTest"