diff --git a/pdf-umbenenner-adapter-out/src/test/java/de/gecheckt/pdf/umbenenner/adapter/outbound/pdfextraction/PdfTextExtractionPortAdapterTest.java b/pdf-umbenenner-adapter-out/src/test/java/de/gecheckt/pdf/umbenenner/adapter/outbound/pdfextraction/PdfTextExtractionPortAdapterTest.java
index f1d5927..b7387ea 100644
--- a/pdf-umbenenner-adapter-out/src/test/java/de/gecheckt/pdf/umbenenner/adapter/outbound/pdfextraction/PdfTextExtractionPortAdapterTest.java
+++ b/pdf-umbenenner-adapter-out/src/test/java/de/gecheckt/pdf/umbenenner/adapter/outbound/pdfextraction/PdfTextExtractionPortAdapterTest.java
@@ -15,6 +15,7 @@ import java.nio.file.Files;
 import java.nio.file.Path;
 
 import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertFalse;
 import static org.junit.jupiter.api.Assertions.assertInstanceOf;
 import static org.junit.jupiter.api.Assertions.assertNotNull;
 import static org.junit.jupiter.api.Assertions.assertThrows;
@@ -125,6 +126,28 @@ class PdfTextExtractionPortAdapterTest {
         assertNotNull(success.extractedText());  // May be empty, but not null
     }
 
+    @Test
+    void testCorruptedPdfReturnsTechnicalError() throws Exception {
+        // Create a file with binary garbage content – exists on disk but is not a valid PDF
+        Path corruptFile = tempDir.resolve("corrupt.pdf");
+        Files.write(corruptFile, "THIS IS NOT A PDF - RANDOM GARBAGE XYZ123!@#".getBytes());
+
+        SourceDocumentCandidate candidate = new SourceDocumentCandidate(
+            "corrupt.pdf",
+            Files.size(corruptFile),
+            new SourceDocumentLocator(corruptFile.toAbsolutePath().toString())
+        );
+
+        PdfExtractionResult result = adapter.extractTextAndPageCount(candidate);
+
+        // PDFBox cannot load a garbage file; the adapter must catch this and return TechnicalError
+        assertInstanceOf(PdfExtractionTechnicalError.class, result,
+            "Binary garbage file (not a real PDF) must yield TechnicalError, not Success");
+        PdfExtractionTechnicalError error = (PdfExtractionTechnicalError) result;
+        assertNotNull(error.errorMessage(), "TechnicalError must carry a non-null error message");
+        assertFalse(error.errorMessage().isBlank(), "TechnicalError message must not be blank");
+    }
+
     // --- Helper methods to create test PDFs ---
 
     /**
diff --git a/pdf-umbenenner-application/src/test/java/de/gecheckt/pdf/umbenenner/application/usecase/M2BatchRunProcessingUseCaseTest.java b/pdf-umbenenner-application/src/test/java/de/gecheckt/pdf/umbenenner/application/usecase/M2BatchRunProcessingUseCaseTest.java
index dd2d560..8f344ed 100644
--- a/pdf-umbenenner-application/src/test/java/de/gecheckt/pdf/umbenenner/application/usecase/M2BatchRunProcessingUseCaseTest.java
+++ b/pdf-umbenenner-application/src/test/java/de/gecheckt/pdf/umbenenner/application/usecase/M2BatchRunProcessingUseCaseTest.java
@@ -24,7 +24,9 @@ import java.net.URI;
 import java.nio.file.Files;
 import java.nio.file.Path;
 import java.time.Instant;
+import java.util.LinkedHashMap;
 import java.util.List;
+import java.util.Map;
 
 import static org.junit.jupiter.api.Assertions.*;
 
@@ -258,6 +260,45 @@ class M2BatchRunProcessingUseCaseTest {
         assertTrue(lockPort.wasReleaseCalled(), "Lock should be released even when source access fails");
     }
 
+    /**
+     * Mixed-batch test: one document per M3 outcome type in a single run.
+     * Proves that no individual outcome aborts the overall batch (AP-008 explicit contract).
+     */
+    @Test
+    void execute_m3MixedBatch_allOutcomeTypes_batchOverallSucceeds() throws Exception {
+        MockRunLockPort lockPort = new MockRunLockPort();
+        // maxPages=3 in buildConfig; pageLimitCandidate has 10 pages → exceeds limit
+        StartConfiguration config = buildConfig(tempDir);
+
+        SourceDocumentCandidate goodCandidate          = makeCandidate("good.pdf");
+        SourceDocumentCandidate noTextCandidate        = makeCandidate("notext.pdf");
+        SourceDocumentCandidate pageLimitCandidate     = makeCandidate("toobig.pdf");
+        SourceDocumentCandidate technicalErrorCandidate = makeCandidate("broken.pdf");
+        SourceDocumentCandidate contentErrorCandidate  = makeCandidate("encrypted.pdf");
+
+        FixedCandidatesPort candidatesPort = new FixedCandidatesPort(List.of(
+                goodCandidate, noTextCandidate, pageLimitCandidate,
+                technicalErrorCandidate, contentErrorCandidate));
+
+        MappedExtractionPort extractionPort = new MappedExtractionPort()
+                .with(goodCandidate,           new PdfExtractionSuccess("Invoice text", new PdfPageCount(1)))
+                .with(noTextCandidate,         new PdfExtractionSuccess("   ", new PdfPageCount(1)))
+                .with(pageLimitCandidate,      new PdfExtractionSuccess("Some text", new PdfPageCount(10)))
+                .with(technicalErrorCandidate, new PdfExtractionTechnicalError("I/O error", null))
+                .with(contentErrorCandidate,   new PdfExtractionContentError("PDF is encrypted"));
+
+        M2BatchRunProcessingUseCase useCase = new M2BatchRunProcessingUseCase(
+                config, lockPort, candidatesPort, extractionPort);
+        BatchRunContext context = new BatchRunContext(new RunId("m3-mixed"), Instant.now());
+
+        BatchRunOutcome outcome = useCase.execute(context);
+
+        assertTrue(outcome.isSuccess(),
+                "Mixed batch with all M3 outcome types must yield batch SUCCESS");
+        assertEquals(5, extractionPort.callCount(),
+                "Extraction must be attempted for each of the 5 candidates");
+    }
+
     @Test
     void execute_m3MultipleCandidates_allProcessed_batchSucceeds() throws Exception {
         MockRunLockPort lockPort = new MockRunLockPort();
@@ -416,4 +457,27 @@ class M2BatchRunProcessingUseCaseTest {
             throw new UnsupportedOperationException("Should not be called");
         }
     }
+
+    /** Per-candidate extraction port: maps each candidate to a fixed result; counts calls. */
+    private static class MappedExtractionPort implements PdfTextExtractionPort {
+        private final Map<SourceDocumentCandidate, PdfExtractionResult> resultMap = new LinkedHashMap<>();
+        private int calls = 0;
+
+        MappedExtractionPort with(SourceDocumentCandidate candidate, PdfExtractionResult result) {
+            resultMap.put(candidate, result);
+            return this;
+        }
+
+        @Override
+        public PdfExtractionResult extractTextAndPageCount(SourceDocumentCandidate candidate) {
+            calls++;
+            PdfExtractionResult result = resultMap.get(candidate);
+            if (result == null) {
+                throw new IllegalStateException("No extraction result mapped for candidate: " + candidate);
+            }
+            return result;
+        }
+
+        int callCount() { return calls; }
+    }
 }