diff --git a/.gitattributes b/.gitattributes
new file mode 100644
index 000000000..484fd26c8
--- /dev/null
+++ b/.gitattributes
@@ -0,0 +1,22 @@
+# Alef-generated binding and e2e files — collapsed in GitHub PR diffs
+packages/csharp/**        linguist-generated=true
+packages/dart/**          linguist-generated=true
+packages/elixir/**        linguist-generated=true
+packages/gleam/**         linguist-generated=true
+packages/go/**            linguist-generated=true
+packages/java/**          linguist-generated=true
+packages/kotlin/**        linguist-generated=true
+packages/php/**           linguist-generated=true
+packages/python/**        linguist-generated=true
+packages/r/**             linguist-generated=true
+packages/ruby/**          linguist-generated=true
+packages/swift/**         linguist-generated=true
+packages/typescript/**    linguist-generated=true
+packages/wasm/**          linguist-generated=true
+packages/zig/**           linguist-generated=true
+crates/kreuzberg-ffi/**   linguist-generated=true
+crates/kreuzberg-node/**  linguist-generated=true
+crates/kreuzberg-php/**   linguist-generated=true
+crates/kreuzberg-py/**    linguist-generated=true
+crates/kreuzberg-wasm/**  linguist-generated=true
+e2e/**                    linguist-generated=true
diff --git a/Cargo.lock b/Cargo.lock
index f8b54ce0e..ce7418007 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -224,6 +224,9 @@ name = "arbitrary"
 version = "1.4.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c3d036a3c4ab069c7b410a2ce876bd74808d2d0888a82667669f8e783a898bf1"
+dependencies = [
+ "derive_arbitrary",
+]
 
 [[package]]
 name = "arg_enum_proc_macro"
@@ -780,6 +783,15 @@ dependencies = [
  "serde",
 ]
 
+[[package]]
+name = "bzip2"
+version = "0.5.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "49ecfb22d906f800d4fe833b6282cf4dc1c298f5057ca0b5445e5c209735ca47"
+dependencies = [
+ "bzip2-sys",
+]
+
 [[package]]
 name = "bzip2"
 version = "0.6.1"
@@ -789,6 +801,16 @@ dependencies = [
  "libbz2-rs-sys",
 ]
 
+[[package]]
+name = "bzip2-sys"
+version = "0.1.13+1.0.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "225bff33b2141874fe80d71e07d6eec4f85c5c216453dd96388240f96e1acc14"
+dependencies = [
+ "cc",
+ "pkg-config",
+]
+
 [[package]]
 name = "calamine"
 version = "0.34.0"
@@ -931,6 +953,17 @@ dependencies = [
  "uuid",
 ]
 
+[[package]]
+name = "cfb"
+version = "0.10.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d8a4f8e55be323b378facfcf1f06aa97f6ec17cf4ac84fb17325093aaf62da41"
+dependencies = [
+ "byteorder",
+ "fnv",
+ "uuid",
+]
+
 [[package]]
 name = "cfb"
 version = "0.14.0"
@@ -1734,6 +1767,17 @@ dependencies = [
  "powerfmt",
 ]
 
+[[package]]
+name = "derive_arbitrary"
+version = "1.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1e567bd82dcff979e4b03460c307b3cdc9e96fde3d73bed1496d2bc75d9dd62a"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.117",
+]
+
 [[package]]
 name = "derive_builder"
 version = "0.20.2"
@@ -3632,6 +3676,7 @@ dependencies = [
  "tracing-opentelemetry",
  "tracing-subscriber",
  "tree-sitter-language-pack",
+ "unhwp",
  "unicode-normalization",
  "ureq 3.3.0",
  "urlencoding",
@@ -3998,6 +4043,16 @@ version = "0.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "112b39cec0b298b6c1999fee3e31427f74f676e4cb9879ed1a121b43661a4154"
 
+[[package]]
+name = "lzma-rs"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "297e814c836ae64db86b36cf2a557ba54368d03f6afcd7d947c266692f71115e"
+dependencies = [
+ "byteorder",
+ "crc",
+]
+
 [[package]]
 name = "lzma-rust2"
 version = "0.15.7"
@@ -4014,6 +4069,17 @@ version = "0.16.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "47bb1e988e6fb779cf720ad431242d3f03167c1b3f2b1aae7f1a94b2495b36ae"
 
+[[package]]
+name = "lzma-sys"
+version = "0.1.20"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5fda04ab3764e6cde78b9974eec4f779acaba7c4e84b36eca3cf77c581b85d27"
+dependencies = [
+ "cc",
+ "libc",
+ "pkg-config",
+]
+
 [[package]]
 name = "lzw"
 version = "0.10.0"
@@ -5133,6 +5199,17 @@ dependencies = [
  "unicase",
 ]
 
+[[package]]
+name = "pulldown-cmark"
+version = "0.12.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f86ba2052aebccc42cbbb3ed234b8b13ce76f75c3551a303cb2bcffcff12bb14"
+dependencies = [
+ "bitflags 2.11.1",
+ "memchr",
+ "unicase",
+]
+
 [[package]]
 name = "pulldown-cmark"
 version = "0.13.3"
@@ -5152,6 +5229,15 @@ version = "0.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "007d8adb5ddab6f8e3f491ac63566a7d5002cc7ed73901f72057943fa71ae1ae"
 
+[[package]]
+name = "pulldown-cmark-to-cmark"
+version = "18.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1e02b63adcb49f2eb675b1694b413b3e9fedbf549dfe2cc98727ad97a0c30650"
+dependencies = [
+ "pulldown-cmark 0.12.2",
+]
+
 [[package]]
 name = "pxfm"
 version = "0.1.29"
@@ -6163,7 +6249,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "29225600349ef74beda5a9fffb36ac660a24613c0bde9315d0c49be1d51e9c24"
 dependencies = [
  "aes 0.8.4",
- "bzip2",
+ "bzip2 0.6.1",
  "cbc 0.1.2",
  "crc32fast",
  "getrandom 0.4.2",
@@ -7279,6 +7365,26 @@ version = "1.20.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "40ce102ab67701b8526c123c1bab5cbe42d7040ccfd0f64af1a385808d2f43de"
 
+[[package]]
+name = "unhwp"
+version = "0.2.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a7e7d0d14b2b326473082e9c3c8ea8c291b0b998b49ba395de374db9dcaa1b8a"
+dependencies = [
+ "cfb 0.10.0",
+ "flate2",
+ "pulldown-cmark 0.12.2",
+ "pulldown-cmark-to-cmark",
+ "quick-xml 0.37.5",
+ "rayon",
+ "regex",
+ "serde",
+ "serde_json",
+ "thiserror",
+ "unicode-normalization",
+ "zip 2.4.2",
+]
+
 [[package]]
 name = "unicase"
 version = "2.9.0"
@@ -8308,6 +8414,15 @@ dependencies = [
  "rustix",
 ]
 
+[[package]]
+name = "xz2"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "388c44dc09d76f1536602ead6d325eb532f5c122f17782bd57fb47baeeb767e2"
+dependencies = [
+ "lzma-sys",
+]
+
 [[package]]
 name = "y4m"
 version = "0.8.0"
@@ -8433,6 +8548,36 @@ dependencies = [
  "syn 2.0.117",
 ]
 
+[[package]]
+name = "zip"
+version = "2.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fabe6324e908f85a1c52063ce7aa26b68dcb7eb6dbc83a2d148403c9bc3eba50"
+dependencies = [
+ "aes 0.8.4",
+ "arbitrary",
+ "bzip2 0.5.2",
+ "constant_time_eq 0.3.1",
+ "crc32fast",
+ "crossbeam-utils",
+ "deflate64",
+ "displaydoc",
+ "flate2",
+ "getrandom 0.3.4",
+ "hmac",
+ "indexmap",
+ "lzma-rs",
+ "memchr",
+ "pbkdf2",
+ "sha1",
+ "thiserror",
+ "time",
+ "xz2",
+ "zeroize",
+ "zopfli",
+ "zstd",
+]
+
 [[package]]
 name = "zip"
 version = "7.2.0"
@@ -8440,7 +8585,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c42e33efc22a0650c311c2ef19115ce232583abbe80850bc8b66509ebef02de0"
 dependencies = [
  "aes 0.8.4",
- "bzip2",
+ "bzip2 0.6.1",
  "constant_time_eq 0.3.1",
  "crc32fast",
  "deflate64",
diff --git a/alef.toml b/alef.toml
index addc8dcb0..cb4b3b6fb 100644
--- a/alef.toml
+++ b/alef.toml
@@ -560,6 +560,7 @@ types = [
   "GzipExtractor",
   "HtmlExtractor",
   "HwpExtractor",
+  "HwpxExtractor",
   "ImageExtractor",
   "JatsExtractor",
   "JupyterExtractor",
diff --git a/crates/kreuzberg/Cargo.toml b/crates/kreuzberg/Cargo.toml
index 68c99642c..89dae2252 100644
--- a/crates/kreuzberg/Cargo.toml
+++ b/crates/kreuzberg/Cargo.toml
@@ -67,6 +67,7 @@ office = [
     "html",
 ]
 hwp = ["dep:cfb", "dep:flate2"]
+hwpx = ["dep:unhwp"]
 iwork = ["dep:zip", "dep:snap"]
 email = ["dep:mail-parser", "dep:cfb", "dep:outlook-pst", "dep:tempfile", "dep:chrono"]
 html = ["dep:html-to-markdown-rs", "dep:v_htmlescape"]
@@ -208,6 +209,7 @@ formats = [
     "excel",
     "office",
     "hwp",
+    "hwpx",
     "iwork",
     "email",
     "html",
@@ -369,6 +371,7 @@ tower-http = { version = "0.6", features = [
 ], optional = true }
 tracing = { workspace = true }
 tracing-opentelemetry = { version = "0.32", optional = true }
+unhwp = { version = "0.2.4", default-features = false, features = ["hwpx"], optional = true }
 unicode-normalization = { version = "0.1.25", optional = true }
 urlencoding = "2"
 utoipa = { version = "5.4", features = ["axum_extras"], optional = true }
diff --git a/crates/kreuzberg/src/core/mime.rs b/crates/kreuzberg/src/core/mime.rs
index 069a9c647..9d27a8bab 100644
--- a/crates/kreuzberg/src/core/mime.rs
+++ b/crates/kreuzberg/src/core/mime.rs
@@ -41,6 +41,7 @@ pub(crate) const SOURCE_CODE_MIME_TYPE: &str = "text/x-source-code";
 
 pub(crate) const EXCEL_MIME_TYPE: &str = "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet";
 
+pub(crate) const HWPX_MIME_TYPE: &str = "application/haansofthwpx";
 pub(crate) const IWORK_PAGES_MIME_TYPE: &str = "application/x-iwork-pages-sffpages";
 pub(crate) const IWORK_NUMBERS_MIME_TYPE: &str = "application/x-iwork-numbers-sffnumbers";
 pub(crate) const IWORK_KEYNOTE_MIME_TYPE: &str = "application/x-iwork-keynote-sffkey";
@@ -783,6 +784,12 @@ fn detect_office_format_from_zip(content: &[u8]) -> Option<&'static str> {
     const NUMBERS_MARKER: &[u8] = b"Index/CalculationEngine.iwa";
     const KEYNOTE_MARKER: &[u8] = b"Index/Presentation.iwa";
 
+    // HWPX: ZIP of OWPML XML, contains Contents/content.hpf manifest
+    const HWPX_MARKER: &[u8] = b"Contents/content.hpf";
+    if contains_subsequence(content, HWPX_MARKER) {
+        return Some(HWPX_MIME_TYPE);
+    }
+
     // Check iWork first (before generic Office) since iWork ZIPs also contain XML
     if contains_subsequence(content, PAGES_MARKER) {
         return Some(IWORK_PAGES_MIME_TYPE);
diff --git a/crates/kreuzberg/src/extractors/hwp.rs b/crates/kreuzberg/src/extractors/hwp.rs
index 0a4558b08..37f43b9c2 100644
--- a/crates/kreuzberg/src/extractors/hwp.rs
+++ b/crates/kreuzberg/src/extractors/hwp.rs
@@ -88,7 +88,7 @@ impl DocumentExtractor for HwpExtractor {
     }
 
     fn supported_mime_types(&self) -> &[&str] {
-        &["application/x-hwp", "application/haansofthwpx"]
+        &["application/x-hwp"]
     }
 
     fn priority(&self) -> i32 {
@@ -106,10 +106,7 @@ mod tests {
         assert_eq!(extractor.name(), "hwp-extractor");
         assert_eq!(extractor.version(), env!("CARGO_PKG_VERSION"));
         assert_eq!(extractor.priority(), 50);
-        assert_eq!(
-            extractor.supported_mime_types(),
-            &["application/x-hwp", "application/haansofthwpx"]
-        );
+        assert_eq!(extractor.supported_mime_types(), &["application/x-hwp"]);
     }
 
     #[test]
@@ -118,4 +115,20 @@ mod tests {
         assert!(extractor.initialize().is_ok());
         assert!(extractor.shutdown().is_ok());
     }
+
+    #[test]
+    fn test_hwpx_mime_not_routed_to_hwp_extractor() {
+        use crate::KreuzbergError;
+        use crate::plugins::registry::DocumentExtractorRegistry;
+        use std::sync::Arc;
+
+        let mut registry = DocumentExtractorRegistry::new();
+        registry.register(Arc::new(HwpExtractor::new())).unwrap();
+
+        let result = registry.get("application/haansofthwpx");
+        assert!(
+            matches!(result, Err(KreuzbergError::UnsupportedFormat(_))),
+            "application/haansofthwpx must not be routed to HwpExtractor"
+        );
+    }
 }
diff --git a/crates/kreuzberg/src/extractors/hwpx.rs b/crates/kreuzberg/src/extractors/hwpx.rs
new file mode 100644
index 000000000..45a8449a9
--- /dev/null
+++ b/crates/kreuzberg/src/extractors/hwpx.rs
@@ -0,0 +1,192 @@
+//! Hangul Word Processor XML (.hwpx) extractor.
+//!
+//! Extracts text, headings, tables, and images from HWPX documents using the `unhwp` crate.
+
+use std::borrow::Cow;
+
+use async_trait::async_trait;
+use bytes::Bytes;
+
+use crate::Result;
+use crate::core::config::ExtractionConfig;
+use crate::plugins::{DocumentExtractor, Plugin};
+use crate::types::ExtractedImage;
+use crate::types::internal::InternalDocument;
+use crate::types::internal_builder::InternalDocumentBuilder;
+
+/// Extractor for Hangul Word Processor XML (.hwpx) files.
+///
+/// Supports HWPX (Open HWPML), the ZIP-based XML successor to the binary HWP 5.0 format.
+pub struct HwpxExtractor;
+
+impl HwpxExtractor {
+    pub(crate) fn new() -> Self {
+        Self
+    }
+}
+
+impl Default for HwpxExtractor {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+impl Plugin for HwpxExtractor {
+    fn name(&self) -> &str {
+        "hwpx-extractor"
+    }
+
+    fn version(&self) -> String {
+        env!("CARGO_PKG_VERSION").to_string()
+    }
+
+    fn initialize(&self) -> Result<()> {
+        Ok(())
+    }
+
+    fn shutdown(&self) -> Result<()> {
+        Ok(())
+    }
+
+    fn description(&self) -> &str {
+        "Hangul Word Processor XML (.hwpx) text extraction"
+    }
+
+    fn author(&self) -> &str {
+        "Kreuzberg Team"
+    }
+}
+
+fn mime_to_format(mime: &str) -> Cow<'static, str> {
+    match mime {
+        "image/png" => Cow::Borrowed("png"),
+        "image/jpeg" | "image/jpg" => Cow::Borrowed("jpeg"),
+        "image/gif" => Cow::Borrowed("gif"),
+        "image/bmp" => Cow::Borrowed("bmp"),
+        "image/webp" => Cow::Borrowed("webp"),
+        _ => Cow::Borrowed("bin"),
+    }
+}
+
+fn build_hwpx_internal_document(doc: unhwp::model::Document, mime_type: &str) -> InternalDocument {
+    let mut builder = InternalDocumentBuilder::new("hwpx");
+    builder.set_mime_type(Cow::Owned(mime_type.to_string()));
+    let mut image_index: usize = 0;
+
+    for section in &doc.sections {
+        for block in &section.content {
+            match block {
+                unhwp::model::Block::Paragraph(p) => {
+                    if p.style.is_heading() && p.has_text_content() {
+                        let text = p.plain_text();
+                        let trimmed = text.trim();
+                        if !trimmed.is_empty() {
+                            builder.push_heading(p.style.heading_level, trimmed, None, None);
+                        }
+                    } else if p.has_text_content() {
+                        let text = p.plain_text();
+                        let trimmed = text.trim();
+                        if !trimmed.is_empty() {
+                            builder.push_paragraph(trimmed, vec![], None, None);
+                        }
+                    }
+
+                    for inline in &p.content {
+                        if let unhwp::model::InlineContent::Image(img_ref) = inline
+                            && let Some(resource) = doc.resources.get(&img_ref.id)
+                        {
+                            let image = ExtractedImage {
+                                data: Bytes::from(resource.data.clone()),
+                                format: mime_to_format(resource.mime_type.as_deref().unwrap_or("")),
+                                image_index,
+                                page_number: None,
+                                width: img_ref.width,
+                                height: img_ref.height,
+                                colorspace: None,
+                                bits_per_component: None,
+                                is_mask: false,
+                                description: img_ref.alt_text.clone(),
+                                ocr_result: None,
+                                bounding_box: None,
+                                source_path: None,
+                                image_kind: None,
+                                kind_confidence: None,
+                                cluster_id: None,
+                            };
+                            builder.push_image(img_ref.alt_text.as_deref(), image, None, None);
+                            image_index += 1;
+                        }
+                    }
+                }
+                unhwp::model::Block::Table(t) => {
+                    if !t.rows.is_empty() {
+                        let cells: Vec<Vec<String>> = t
+                            .rows
+                            .iter()
+                            .map(|row| row.cells.iter().map(|cell| cell.plain_text()).collect())
+                            .collect();
+                        builder.push_table_from_cells(&cells, None, None);
+                    }
+                }
+            }
+        }
+    }
+
+    builder.build()
+}
+
+#[cfg_attr(not(target_arch = "wasm32"), async_trait)]
+#[cfg_attr(target_arch = "wasm32", async_trait(?Send))]
+impl DocumentExtractor for HwpxExtractor {
+    async fn extract_bytes(
+        &self,
+        content: &[u8],
+        mime_type: &str,
+        _config: &ExtractionConfig,
+    ) -> Result<InternalDocument> {
+        let doc = unhwp::parse_bytes(content)
+            .map_err(|e| crate::KreuzbergError::parsing(format!("Failed to parse HWPX: {e}")))?;
+        Ok(build_hwpx_internal_document(doc, mime_type))
+    }
+
+    fn supported_mime_types(&self) -> &[&str] {
+        &["application/haansofthwpx"]
+    }
+
+    fn priority(&self) -> i32 {
+        50
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_hwpx_extractor_plugin_interface() {
+        let extractor = HwpxExtractor::new();
+        assert_eq!(extractor.name(), "hwpx-extractor");
+        assert_eq!(extractor.version(), env!("CARGO_PKG_VERSION"));
+        assert_eq!(extractor.priority(), 50);
+        assert_eq!(extractor.supported_mime_types(), &["application/haansofthwpx"]);
+    }
+
+    #[test]
+    fn test_hwpx_extractor_initialize_shutdown() {
+        let extractor = HwpxExtractor::new();
+        assert!(extractor.initialize().is_ok());
+        assert!(extractor.shutdown().is_ok());
+    }
+
+    #[cfg(feature = "hwp")]
+    #[test]
+    fn test_hwpx_not_routed_to_hwp_extractor() {
+        use crate::extractors::hwp::HwpExtractor;
+
+        let hwp = HwpExtractor::new();
+        assert!(
+            !hwp.supported_mime_types().contains(&"application/haansofthwpx"),
+            "HwpExtractor must not claim application/haansofthwpx"
+        );
+    }
+}
diff --git a/crates/kreuzberg/src/extractors/mod.rs b/crates/kreuzberg/src/extractors/mod.rs
index f2da8aa22..e6eb15ca7 100644
--- a/crates/kreuzberg/src/extractors/mod.rs
+++ b/crates/kreuzberg/src/extractors/mod.rs
@@ -97,6 +97,9 @@ pub mod excel;
 #[cfg(feature = "hwp")]
 pub mod hwp;
 
+#[cfg(feature = "hwpx")]
+pub mod hwpx;
+
 #[cfg(feature = "iwork")]
 pub mod iwork;
 
@@ -197,6 +200,9 @@ pub use excel::ExcelExtractor;
 #[cfg(feature = "hwp")]
 pub use hwp::HwpExtractor;
 
+#[cfg(feature = "hwpx")]
+pub use hwpx::HwpxExtractor;
+
 #[cfg(feature = "iwork")]
 pub use iwork::{keynote::KeynoteExtractor, numbers::NumbersExtractor, pages::PagesExtractor};
 
@@ -368,6 +374,11 @@ pub(crate) fn register_default_extractors() -> Result<()> {
         registry.register(Arc::new(HwpExtractor::new()))?;
     }
 
+    #[cfg(feature = "hwpx")]
+    {
+        registry.register(Arc::new(HwpxExtractor::new()))?;
+    }
+
     #[cfg(feature = "iwork")]
     {
         registry.register(Arc::new(PagesExtractor::new()))?;
diff --git a/fixtures/format_specific/format_hwpx_standalone.json b/fixtures/format_specific/format_hwpx_standalone.json
new file mode 100644
index 000000000..97d1b7274
--- /dev/null
+++ b/fixtures/format_specific/format_hwpx_standalone.json
@@ -0,0 +1,26 @@
+{
+  "id": "format_hwpx_standalone",
+  "category": "format_specific",
+  "description": "Standalone HWPX extraction using extract_bytes_sync",
+  "tags": ["format_specific", "hwpx", "text_extraction"],
+  "call": "extract_bytes_sync",
+  "input": {
+    "data": "hwpx/simple.hwpx",
+    "mime_type": "application/haansofthwpx"
+  },
+  "assertions": [
+    {
+      "type": "not_error"
+    },
+    {
+      "type": "min_length",
+      "field": "content",
+      "value": 20
+    },
+    {
+      "type": "contains",
+      "field": "content",
+      "value": "Hello from HWPX"
+    }
+  ]
+}
diff --git a/test_documents/hwpx/simple.hwpx b/test_documents/hwpx/simple.hwpx
new file mode 100644
index 000000000..1e8963a34
Binary files /dev/null and b/test_documents/hwpx/simple.hwpx differ