feat: adopt v2 AI tasks with exact QPK source binding

.github/actionlint.yaml

@@ -0,0 +1,3 @@
+self-hosted-runner:
+  labels:
+    - research-runner

.github/workflows/cn-index-etf-research.yml

@@ -0,0 +1,38 @@
+name: CN Index ETF Research V2
+on:
+  workflow_dispatch:
+    inputs:
+      execute:
+        description: Run the root-policy-bound cloud experiment
+        type: boolean
+        default: false
+permissions:
+  contents: read
+  id-token: write
+concurrency:
+  group: cn-index-etf-research-v2
+  cancel-in-progress: false
+jobs:
+  research:
+    if: github.repository == 'QuantStrategyLab/CnEquityStrategies' && github.ref == 'refs/heads/main' && github.run_attempt == 1 && vars.AI_SERVICE_RELEASE_READY == 'true'
+    runs-on: [self-hosted, research-runner]
+    timeout-minutes: 60
+    env:
+      AI_SERVICE_URL: ${{ vars.AI_SERVICE_URL }}
+      AI_SERVICE_AUDIENCE: ${{ vars.AI_SERVICE_AUDIENCE }}
+      AI_SERVICE_MODE: agent
+      AI_SERVICE_MODEL: ${{ vars.AI_SERVICE_RESEARCH_MODEL }}
+      AI_SERVICE_PROFILE: ${{ vars.AI_SERVICE_RESEARCH_PROFILE || 'default' }}
+      AI_SERVICE_SUMMARY_MODEL: ${{ vars.AI_SERVICE_SUMMARY_MODEL }}
+      STRATEGY_WATCH_SOURCE_REPO: QuantStrategyLab/CnEquitySnapshotPipelines
+      GH_TOKEN: ${{ github.token }}
+      WATCHER_RESULT: ${{ vars.CN_RESEARCH_WATCHER_RESULT_PATH }}
+      EXECUTE: ${{ inputs.execute }}
+    steps:
+      - name: Run the preinstalled approved source revision
+        run: |
+          set -euo pipefail
+          test -n "$WATCHER_RESULT"
+          args=(--dry-run)
+          if [[ "$EXECUTE" == "true" ]]; then args=(); fi
+          python -m cn_equity_strategies.backtest.index_etf_research_task --watcher-result "$WATCHER_RESULT" "${args[@]}"

.github/workflows/drift-check.yml

@@ -194,17 +194,18 @@ jobs:
       contents: read
       issues: write
       id-token: write
-    uses: QuantStrategyLab/QuantPlatformKit/.github/workflows/reusable-drift-check.yml@c7646a7168b3dafa763ef7751a182d23e8de7790
+    uses: ./.github/workflows/reusable-drift-check.yml
     with:
       strategy_domain: cn_equity
       caller_event_name: ${{ github.event_name }}
       caller_pr_head_repository: ${{ github.event.pull_request.head.repo.full_name || '' }}
       snapshot_repository: QuantStrategyLab/CnEquitySnapshotPipelines
       snapshot_checkout_path: external/CnEquitySnapshotPipelines
       snapshot_repository_ref: ${{ needs.preflight_backtests.outputs.snapshot_repository_ref }}
-      ai_gateway_service_url: ${{ vars.AI_GATEWAY_SERVICE_URL }}
+      ai_service_url: ${{ vars.AI_SERVICE_URL }}
+      ai_service_audience: ${{ vars.AI_SERVICE_AUDIENCE }}
+      ai_service_reviewers_json: ${{ vars.AI_SERVICE_REVIEWERS_JSON }}
       quant_platform_kit_ref: c7646a7168b3dafa763ef7751a182d23e8de7790
       lifecycle_preflight_artifact: lifecycle-preflight-${{ github.run_id }}-${{ github.run_attempt }}
     secrets:
-      codex_audit_service_url: ${{ secrets.CODEX_AUDIT_SERVICE_URL }}
       snapshot_repository_token: ${{ secrets.SNAPSHOT_REPOSITORY_TOKEN || secrets.QSL_REPO_SYNC_TOKEN || github.token }}

.github/workflows/evidence-gate.yml

@@ -70,17 +70,10 @@ jobs:
           python -m pip install -e ".[research]" pandas
           python -m pip check
 
-      - name: Checkout AIAuditBridge
-        uses: actions/checkout@v6
-        with:
-          repository: QuantStrategyLab/AIAuditBridge
-          ref: 60bd64a2ae059a082614181eeb845b46df395523
-          path: external/AIAuditBridge
-
       - name: Evaluate Evidence Gate
         env:
           GITHUB_BASE_REF: ${{ github.base_ref }}
-          AIAUDIT_BRIDGE_ROOT: external/AIAuditBridge
-          CODEX_AUDIT_SERVICE_URL: ${{ secrets.CODEX_AUDIT_SERVICE_URL }}
-          AI_GATEWAY_SERVICE_URL: ${{ vars.AI_GATEWAY_SERVICE_URL }}
+          AI_SERVICE_URL: ${{ vars.AI_SERVICE_URL }}
+          AI_SERVICE_AUDIENCE: ${{ vars.AI_SERVICE_AUDIENCE }}
+          AI_SERVICE_REVIEWERS_JSON: ${{ vars.AI_SERVICE_REVIEWERS_JSON }}
         run: python scripts/gate_evidence_package.py