diff --git a/.github/copilot-code-review.md b/.github/copilot-code-review.md
new file mode 100644
index 00000000..94476f63
--- /dev/null
+++ b/.github/copilot-code-review.md
@@ -0,0 +1,14 @@
+You are a rigorous senior code reviewer tasked with preventing security vulnerabilities in code submissions.
+Your assessment must be based on the code diffs of each commit.
+
+- Language: English
+- Focus on .NET security policy and best practices
+- Flag any potential SQL injection, XSS, path traversal, insecure deserialization, or other OWASP Top 10 risks
+- Check for hardcoded secrets, credentials, or sensitive data exposure in application source code
+- Verify proper input validation and output encoding
+- Ensure secure file I/O patterns (no arbitrary file access)
+
+IMPORTANT: Do NOT flag the following as security issues:
+- Using ${{ secrets.* }} in GitHub Actions workflows (this is the correct way to use secrets in CI)
+- Changes to CI/CD configuration files (.yml/.yaml under .github/workflows/) unless they contain actual hardcoded credentials
+- Changes to documentation files (.md) unless they expose sensitive information
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index c210469f..2d153f70 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -8,7 +8,7 @@ on:
 
 permissions:
   contents: read
-  pull-requests: write
+  pull-requests: read
 
 jobs:
   build:
@@ -39,29 +39,99 @@ jobs:
           name: test-results
           path: '**/test-results.trx'
 
-  ai-pr-review:
+  ai-security-scan:
     if: github.event_name == 'pull_request'
     runs-on: ubuntu-latest
     needs: build
 
-    permissions:
-      contents: read
-      pull-requests: write
-
     steps:
       - name: Checkout
         uses: actions/checkout@v4
-
-      - name: AI PR Review
-        uses: github/copilot-code-review-action@v1
         with:
-          model: gpt-4o
-          custom_instructions: |
-            You are a rigorous senior code reviewer tasked with preventing security vulnerabilities in code submissions.
-            Your assessment must be based on the code diffs of each commit.
-            - Language: English
-            - Focus on .NET security policy and best practices
-            - Flag any potential SQL injection, XSS, path traversal, insecure deserialization, or other OWASP Top 10 risks
-            - Check for hardcoded secrets, credentials, or sensitive data exposure
-            - Verify proper input validation and output encoding
-            - Ensure secure file I/O patterns (no arbitrary file access)
+          fetch-depth: 0
+
+      - name: Get PR diff
+        id: diff
+        run: |
+          DIFF=$(git diff origin/${{ github.base_ref }}...HEAD -- '*.cs' '*.csproj' || true)
+          if [ -z "$DIFF" ]; then
+            echo "No code changes detected."
+            echo "skip=true" >> $GITHUB_OUTPUT
+          else
+            # Save diff to file to avoid shell escaping issues
+            echo "$DIFF" > /tmp/pr_diff.txt
+            echo "skip=false" >> $GITHUB_OUTPUT
+          fi
+
+      - name: AI Security Review
+        if: steps.diff.outputs.skip != 'true'
+        env:
+          AZURE_OPENAI_API_KEY: ${{ secrets.AZURE_OPENAI_API_KEY }}
+          AZURE_OPENAI_ENDPOINT: ${{ secrets.AZURE_OPENAI_ENDPOINT }}
+          AZURE_OPENAI_DEPLOYMENT: ${{ secrets.AZURE_OPENAI_DEPLOYMENT }}
+        run: |
+          DIFF=$(cat /tmp/pr_diff.txt)
+          
+          # Truncate diff if too large (max ~12000 chars to fit in context)
+          if [ ${#DIFF} -gt 12000 ]; then
+            DIFF="${DIFF:0:12000}... [truncated]"
+          fi
+          
+          INSTRUCTIONS=$(cat .github/copilot-code-review.md 2>/dev/null || echo "Review for security issues.")
+          
+          API_VERSION="2025-01-01-preview"
+          URL="${AZURE_OPENAI_ENDPOINT%/}/openai/deployments/${AZURE_OPENAI_DEPLOYMENT}/chat/completions?api-version=${API_VERSION}"
+          
+          # Build JSON payload safely using jq
+          PAYLOAD=$(jq -n \
+            --arg instructions "$INSTRUCTIONS" \
+            --arg diff "$DIFF" \
+            '{
+              messages: [
+                { role: "system", content: $instructions },
+                { role: "user", content: ("Review this code diff for security vulnerabilities in APPLICATION SOURCE CODE ONLY. Respond ONLY with a JSON object (no markdown, no code blocks): {\"passed\": true/false, \"issues\": [\"description1\", \"description2\"]}. Set passed=true if no security issues found in application code, passed=false only for real security concerns like SQL injection, XSS, path traversal, hardcoded credentials in source code, etc. Do NOT flag CI/CD workflow configuration or documentation changes as issues.\n\nDiff:\n" + $diff) }
+              ],
+              temperature: 0.1
+            }')
+          
+          RESPONSE=$(curl -s "$URL" \
+            -H "Content-Type: application/json" \
+            -H "api-key: $AZURE_OPENAI_API_KEY" \
+            -d "$PAYLOAD")
+          
+          # Extract content
+          CONTENT=$(echo "$RESPONSE" | jq -r '.choices[0].message.content // empty')
+          
+          if [ -z "$CONTENT" ]; then
+            echo "::error::Failed to get AI review response"
+            echo "$RESPONSE" | jq .
+            exit 1
+          fi
+          
+          echo "=== AI Security Review Result ==="
+          echo "$CONTENT"
+          echo "================================="
+          
+          # Parse JSON from response: strip markdown code blocks, then parse with jq
+          CLEAN_CONTENT=$(echo "$CONTENT" | sed '/^```/d')
+          # Use 'if .passed then "true" else "false" end' to handle boolean false correctly
+          PASSED=$(echo "$CLEAN_CONTENT" | jq -r 'if .passed == true then "true" elif .passed == false then "false" else "unknown" end' 2>/dev/null)
+          
+          if [ "$PASSED" = "false" ]; then
+            echo ""
+            echo "::error::AI Security Review FAILED - security issues detected"
+            echo "$CLEAN_CONTENT" | jq -r '.issues[]?' 2>/dev/null | while read -r issue; do
+              echo "::warning::$issue"
+            done
+            exit 1
+          elif [ "$PASSED" = "true" ]; then
+            echo ""
+            echo "✅ AI Security Review PASSED - no security issues found"
+          else
+            echo "::warning::Could not parse AI review result, treating as FAIL for safety"
+            exit 1
+          fi
+
+      - name: Skip notice
+        if: steps.diff.outputs.skip == 'true'
+        run: echo "✅ No code changes to review"
diff --git a/.gitignore b/.gitignore
index 6752c8bc..e3258262 100644
--- a/.gitignore
+++ b/.gitignore
@@ -14,3 +14,6 @@ Desktop.ini
 
 ## NuGet
 *.nupkg
+
+## Wiki
+MiniPdf.wiki/
diff --git a/README.md b/README.md
index 99917d03..a172b630 100644
--- a/README.md
+++ b/README.md
@@ -2,6 +2,8 @@
 
 A minimal, zero-dependency .NET library for generating PDF documents from text and Excel (.xlsx) files.
 
+> **Security**: All PRs are automatically reviewed by Copilot AI and Azure AI security scan for vulnerabilities.
+
 ## Features
 
 - **Text-to-PDF** — Create PDF documents with positioned or auto-wrapped text