feat(studio): add migration effect evaluation

frontend/README.md

@@ -68,12 +68,18 @@ server that `veadk frontend` launches — no separate backend.
   action is required.
 - **Existing Agent migration**: upload a local project ZIP for read-only
   analysis, confirm the detected framework and entry point, then migrate and
-  validate it in a temporary Sandbox. Successful migration source is saved as
-  an immutable version in the same private Studio TOS project store. The
-  separate “已迁移项目” page can view, download, deploy, delete, and compare
-  versions; any version can be restored into the intelligent-development flow
-  for another intent-driven iteration after the temporary migration environment
-  has ended.
+  validate it in a temporary Sandbox. An optional migration-effect evaluation
+  is off by default; when enabled, users can enter 1–100 real user questions by
+  hand or bulk paste, while expected outcomes and criteria remain optional.
+  Standard evaluation uses three dimensions; users can instead select custom
+  dimensions before upload. The locked dataset and final HTML report are stored
+  as immutable owner-only TOS assets. The report is fetched and rendered in a
+  side drawer only after the user selects “View report,” and remains available
+  for download. Successful migration source is saved as an
+  immutable version in the same private Studio TOS project store. The separate
+  “已迁移项目” page can view, download, deploy, delete, and compare versions;
+  any version can be restored into the intelligent-development flow for another
+  intent-driven iteration after the temporary migration environment has ended.
 - **Reasoning & tool calls** shown inline (collapsible "thinking", tool blocks).
 - **Agent context rail** keeps the selected Agent's description, model, tools,
   skills, and optional live multi-Agent topology together in the conversation's
@@ -261,19 +267,22 @@ server that `veadk frontend` launches — no separate backend.
   creation, and service publishing as separate deployment stages.
 - **Existing-project migration**: upload one local ZIP of at most 20 MiB from
   the add-Agent menu. Studio creates one user-owned Dev Sandbox Session with a
-  one-hour TTL, then asks the preinstalled Codex to perform read-only framework,
-  entry-point, and migration-boundary analysis. Migration starts only after the
-  user confirms the framework, entry point, and open questions. Structured
-  frameworks run the preinstalled `ak migrate`; Dify and Any projects run
-  `ak migrate --execution in-place` with Codex in the same Session. State,
-  logs, and artifacts remain only under
-  `/home/gem/.studio/migration/v1/` in that Session. Preview, download, and
-  Runtime deployment stop when the Session expires. Runtime deployment resolves
-  and verifies the owned Session artifact on the server instead of trusting
-  browser-provided files or entry points. AgentKit CLI `0.51.1` is only the
-  current baseline; these CLI changes must be released as a new version. The
-  Dev Sandbox image must pin that migration-capable release and its SHA256 at
-  image build time.
+  one-hour TTL, extended to two hours when effect evaluation is enabled, then
+  asks the preinstalled Codex to perform read-only framework, entry-point, and
+  migration-boundary analysis. Migration starts only after the user confirms
+  the framework, entry point, and open questions. Structured frameworks run the
+  preinstalled `ak migrate`; Dify and Any projects run
+  `ak migrate --execution in-place` with Codex in the same Session. Evaluation
+  deploys a temporary Runtime, checkpoints per-case execution as JSONL, judges
+  batches in one fresh resumable Codex thread, and always reconciles Runtime
+  cleanup before completing or cancelling. Reports show 0–100 display scores,
+  execution success, evidence coverage, N/A counts, low-scoring and failed
+  cases, versions, evidence severity, and cleanup status without a pass/fail
+  verdict. Evaluation failure never hides or rolls back the migration artifact.
+  Runtime deployment resolves and verifies the owned Session artifact on the
+  server instead of trusting browser-provided files or entry points. The Dev
+  Sandbox image must pin AgentKit CLI `0.52.16` and its SHA256 at image build
+  time.
 - **Built-in code execution**: selecting `代码执行` adds VeADK's `run_code`
   tool to generated Python and reveals the required `AGENTKIT_TOOL_ID` sandbox
   field and optional `AGENTKIT_TOOL_REGION` field below the built-in tool list.

frontend/server/migration/contracts.py

@@ -31,6 +31,7 @@
     is_valid_model_id,
     is_valid_structured_entry,
 )
+from .evaluation.dimensions import EVALUATION_DIMENSION_IDS, STANDARD_DIMENSION_IDS
 
 _MAX_PATH_BYTES = 4 * 1024
 _MAX_PATH_DEPTH = 64
@@ -184,7 +185,7 @@ def validate_migration_request(
             "session_ttl_seconds",
             "created_at",
         },
-        optional={"model_id"},
+        optional={"model_id", "evaluation"},
     )
     if (
         value.get("schema_version") != 1
@@ -201,6 +202,27 @@ def validate_migration_request(
     _text(value.get("instruction"), maximum=_MAX_TEXT_LENGTH)
     if "model_id" in value and not is_valid_model_id(value.get("model_id")):
         raise MigrationContractError("invalid model id")
+    evaluation = value.get("evaluation")
+    if evaluation is not None:
+        if not isinstance(evaluation, dict):
+            raise MigrationContractError("invalid evaluation config")
+        _exact_keys(
+            evaluation,
+            required={"enabled", "preset", "dimensions"},
+        )
+        enabled = evaluation.get("enabled")
+        preset = evaluation.get("preset")
+        dimensions = evaluation.get("dimensions")
+        if (
+            enabled is not True
+            or preset not in {"standard", "custom"}
+            or not isinstance(dimensions, list)
+            or not dimensions
+            or len(set(str(item) for item in dimensions)) != len(dimensions)
+            or any(item not in EVALUATION_DIMENSION_IDS for item in dimensions)
+            or (preset == "standard" and tuple(dimensions) != STANDARD_DIMENSION_IDS)
+        ):
+            raise MigrationContractError("invalid evaluation config")
     created_at = value.get("created_at")
     if isinstance(created_at, str):
         _timestamp_text(created_at)

frontend/server/migration/evaluation/__init__.py

@@ -0,0 +1,35 @@
+# Copyright (c) 2025 Beijing Volcano Engine Technology Co., Ltd. and/or its affiliates.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Migration-effect evaluation for AgentKit Studio."""
+
+from .dimensions import (
+    EVALUATION_DIMENSIONS,
+    STANDARD_DIMENSION_IDS,
+    EvaluationDimensionId,
+)
+from .models import (
+    EvaluationCaseBody,
+    EvaluationDatasetBody,
+    MigrationEvaluationConfig,
+)
+
+__all__ = [
+    "EVALUATION_DIMENSIONS",
+    "STANDARD_DIMENSION_IDS",
+    "EvaluationCaseBody",
+    "EvaluationDatasetBody",
+    "EvaluationDimensionId",
+    "MigrationEvaluationConfig",
+]