Merge branch 'master' into feat/agentic-dashboard-summary

#1799 landed on master, so the two evaluators now register side by side. The
registry set, the parametrized kind list and the trace-linker list each gain both
entries.

The dispatch needed care rather than concatenation: the two elif branches share
the k/agent_id/**lf_kw tail that follows the conflict marker, so keeping both
headers alone would have spliced one argument tail onto two calls and dropped
dashboard_summary's own arguments. Each kind now has its own complete call.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Author: Tomkess

packages/gooddata-eval/scripts/verify_guardrail_refusal_criteria.py

@@ -39,7 +39,9 @@
 
 from dotenv import load_dotenv
 
-load_dotenv("/Users/petertomko/gdc-mic-ai-evaluation/.env")
+# Whatever .env the caller points at, defaulting to the working directory. It used to be an
+# absolute path, which made the script runnable on exactly one machine.
+load_dotenv(os.environ.get("GD_EVAL_ENV_FILE", ".env"))
 
 from gooddata_eval.core.agentic.guardrail import _GUARDRAIL_EVALUATION_STEPS  # noqa: E402
 from gooddata_eval.core.evaluators._guardrail_criteria import (  # noqa: E402

packages/gooddata-eval/src/gooddata_eval/cli/agentic_runner.py

@@ -21,6 +21,7 @@
 from gooddata_eval.core.agentic.metric_skill import evaluate_agentic_metric_skill
 from gooddata_eval.core.agentic.search_tool import evaluate_agentic_search_tool
 from gooddata_eval.core.agentic.visualization import evaluate_agentic_visualization
+from gooddata_eval.core.agentic.what_if import evaluate_agentic_what_if
 from gooddata_eval.core.config import ReasoningEffort
 from gooddata_eval.core.models import AgenticEvalOutcome, CreatedVisualization, DatasetItem
 from gooddata_eval.core.runner import EvalReport, ItemReport
@@ -49,6 +50,7 @@ class _LfKw(TypedDict, total=False):
         "agentic_conversation",
         "agentic_kda_skill",
         "agentic_dashboard_summary",
+        "agentic_what_if",
     }
 )
 
@@ -288,6 +290,17 @@ def _dispatch_agentic(
             agent_id=agent_id,
             **lf_kw,
         )
+    elif kind == "agentic_what_if":
+        return evaluate_agentic_what_if(
+            host=host,
+            token=token,
+            workspace_id=workspace_id,
+            question=item.question,
+            expected_output=eo if isinstance(eo, dict) else {},
+            k=k,
+            agent_id=agent_id,
+            **lf_kw,
+        )
     elif kind == "agentic_conversation":
         fixture_data = eo.get("fixture") or eo if isinstance(eo, dict) else {}
         return evaluate_agentic_conversation(