chore: Fix trusted-scripts path (#7)

* chore: test

* chore: fix trusted-scripts path

Author: Ariel-Rodriguez

.github/workflows/benchmark-dashboard.yml

@@ -53,6 +53,17 @@ jobs:
             ARTIFACT_NAME="benchmark-${{ inputs.provider }}-${{ inputs.model }}-$(date +%Y%m%d-%H%M%S)"
             mv tests/results "tests/${ARTIFACT_NAME}"
           fi
+          # Create docs/benchmarks if it doesn't exist for publish_benchmarks.py
+          mkdir -p docs/benchmarks
+
+      - name: Generate dashboard
+        working-directory: workspace
+        run: |
+          uv run --project tests python3 ci/publish_benchmarks.py \
+            --provider ${{ inputs.provider }} \
+            --model ${{ inputs.model }} \
+            --branch benchmark-history \
+            --no-benchmark
 
       - name: Generate dashboard
         working-directory: workspace
@@ -79,11 +90,19 @@ jobs:
           ref: benchmark-history
           path: benchmark-data
 
-      - name: Verify deployment
+      - name: Copy results to docs
+        run: |
+          mkdir -p workspace/docs/benchmarks
+          cp benchmark-data/docs/benchmarks.json workspace/docs/benchmarks.json 2>/dev/null || true
+          cp benchmark-data/docs/index.html workspace/docs/index.html 2>/dev/null || true
+          # Also copy individual benchmark results if they exist
+          cp -r benchmark-data/docs/benchmarks/*.json workspace/docs/benchmarks/ 2>/dev/null || true
+
+      - name: Commit and push updates
+        working-directory: workspace
         run: |
-          echo "Benchmark data directory contents:"
-          ls -la benchmark-data/docs/
-          echo ""
-          echo "GitHub Pages will be served from benchmark-history branch"
-          echo "Dashboard will be available at:"
-          echo "https://${{ github.repository }}/index.html"
+          git config user.name "GitHub Actions"
+          git config user.email "actions@github.com"
+          git add docs/
+          git commit -m "Update benchmark data" || echo "No changes to commit"
+          git push origin HEAD:benchmark-history

.github/workflows/skill-validation.yml

@@ -71,8 +71,10 @@ jobs:
           ref: main
           path: trusted-scripts
 
-      # Checkout actual PR code to test
+      # Checkout actual PR code to test (to different path to avoid overwriting trusted-scripts)
       - uses: actions/checkout@v4
+        with:
+          path: pr-code
 
       - uses: actions/setup-node@v4
         if: matrix.provider == 'copilot'
@@ -100,10 +102,11 @@ jobs:
           COPILOT_GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           OLLAMA_API_KEY: ${{ secrets.OLLAMA_API_KEY }}
           TRULENS_OTEL_ENDPOINT: ${{ secrets.TRULENS_OTEL_ENDPOINT }}
+        working-directory: pr-code
         run: |
           # Construct arguments array for safety
           ARGS=(--provider "${{ matrix.provider }}" --model "${{ matrix.model }}" --judge --verbose --report --threshold 50)
-          
+
           if [ -n "${{ matrix.extra_args }}" ]; then
             # Split extra_args safely if needed, but for now assuming simple flags
             ARGS+=(${{ matrix.extra_args }})
@@ -116,15 +119,16 @@ jobs:
           fi
 
           echo "Running evaluation..."
-          # Use trusted script with correct project path
-          uv run --project . --frozen trusted-scripts/tests/evaluator.py "${ARGS[@]}"
+          # Run evaluator from trusted-scripts/tests
+          # The working-directory is pr-code, so evaluator.py finds PR code at current directory
+          uv run --project . --frozen ../trusted-scripts/tests/evaluator.py "${ARGS[@]}"
 
       - name: Upload results
         if: always()
         uses: actions/upload-artifact@v4
         with:
           name: ${{ steps.artifact.outputs.name }}
-          path: tests/results/
+          path: pr-code/tests/results/
           retention-days: 1
 
   consolidate:
@@ -141,14 +145,14 @@ jobs:
 
       - uses: actions/download-artifact@v4
         with:
-          path: tests/results/
+          path: pr-code/tests/results/
 
       - name: Consolidate results
-        run: python3 trusted-scripts/ci/consolidate_results.py
+        run: python3 trusted-scripts/ci/consolidate_results.py --results-dir pr-code/tests/results --output-file pr-code/comment.md
 
       - name: Post to PR
         uses: marocchino/sticky-pull-request-comment@v2
-        if: hashFiles('comment.md') != ''
+        if: hashFiles('pr-code/comment.md') != ''
         with:
           number: ${{ github.event.issue.number }}
-          path: comment.md
+          path: pr-code/comment.md

skills/ps-error-handling-design/SKILL.md

@@ -4,7 +4,7 @@ description: Design systems with explicit error handling. Avoid throwing excepti
 severity: WARN
 ---
 
-## Principle
+# Principle
 
 Treat error handling as a first-class design concern, not an afterthought: