fix: TB regression workflow fixes (venv, Modal, warnings) (#1385)

Co-authored-by: Letta Code <noreply@letta.com>

.github/workflows/terminal-bench-regression.yml

@@ -12,6 +12,9 @@ on:
         description: "Max concurrent tasks"
         default: "4"
 
+env:
+  FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: true
+
 jobs:
   regression:
     runs-on: ubuntu-latest
@@ -26,9 +29,14 @@ jobs:
 
       - name: Setup Python + uv
         uses: astral-sh/setup-uv@v6
+        with:
+          enable-cache: false
 
-      - name: Install Harbor
-        run: uv pip install --system "harbor>=0.1.45" "litellm>=1.0.0"
+      - name: Create venv and install deps
+        run: |
+          uv venv .venv
+          source .venv/bin/activate
+          uv pip install "harbor>=0.1.45" "litellm>=1.0.0" "modal>=1.3.5"
 
       - name: Configure Modal
         env:
@@ -44,6 +52,8 @@ jobs:
           ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
           OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
         run: |
+          source .venv/bin/activate
+
           # Build --task-name flags from regression-tasks.txt
           TASK_FLAGS=""
           while IFS= read -r task; do
@@ -85,6 +95,8 @@ jobs:
 
       - name: Setup Python + uv
         uses: astral-sh/setup-uv@v6
+        with:
+          enable-cache: false
 
       - name: Generate report and update GitHub Issue
         env:
@@ -93,7 +105,7 @@ jobs:
           GITHUB_RUN_ID: ${{ github.run_id }}
           GITHUB_SERVER_URL: ${{ github.server_url }}
         run: |
-          python benchmarks/terminal_bench/report.py \
+          uv run python benchmarks/terminal_bench/report.py \
             --results-dir results/ \
             --baseline benchmarks/terminal_bench/baseline.json \
             --repo "${{ github.repository }}"