diff --git a/src/wetgit/pipeline/downloader.py b/src/wetgit/pipeline/downloader.py
new file mode 100644
index 0000000..06e04fe
--- /dev/null
+++ b/src/wetgit/pipeline/downloader.py
@@ -0,0 +1,41 @@
+"""BWB XML downloader.
+
+Downloadt BWB toestand-XML bestanden van de officiele repository.
+"""
+
+from __future__ import annotations
+
+import logging
+from pathlib import Path
+
+import httpx
+
+from wetgit.pipeline.sru_client import SRURecord
+
+logger = logging.getLogger(__name__)
+
+
+def download_xml(record: SRURecord, output_dir: Path) -> Path | None:
+    """Download de XML voor een SRU record.
+
+    Args:
+        record: SRU record met de download-URL.
+        output_dir: Directory om de XML op te slaan.
+
+    Returns:
+        Pad naar het gedownloade bestand, of None bij fout.
+    """
+    xml_path = output_dir / f"{record.bwb_id}.xml"
+
+    if xml_path.exists():
+        return xml_path
+
+    try:
+        resp = httpx.get(record.xml_url, timeout=60, follow_redirects=True)
+        resp.raise_for_status()
+        xml_path.write_bytes(resp.content)
+        logger.debug("Downloaded %s (%d KB)", record.bwb_id, len(resp.content) // 1024)
+        return xml_path
+    except httpx.HTTPError as e:
+        logger.warning("Download failed for %s: %s", record.bwb_id, e)
+        return None
diff --git a/src/wetgit/pipeline/runner.py b/src/wetgit/pipeline/runner.py
new file mode 100644
index 0000000..3b04a34
--- /dev/null
+++ b/src/wetgit/pipeline/runner.py
@@ -0,0 +1,151 @@
+"""Pipeline runner — orkestreert SRU crawl, download en parse.
+
+Usage:
+    python -m wetgit.pipeline.runner --type wet --limit 100 --output /tmp/wetgit-output
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import time
+from pathlib import Path
+
+from wetgit.pipeline.bwb_parser import parse_bwb_xml
+from wetgit.pipeline.downloader import download_xml
+from wetgit.pipeline.sru_client import BWB_TYPES, SRURecord, fetch_catalogue
+
+logger = logging.getLogger(__name__)
+
+# Mappenstructuur conform PRD
+TYPE_TO_DIR = {
+    "wet": "wet",
+    "AMvB": "amvb",
+    "ministeriele-regeling": "ministeriele-regeling",
+    "KB": "kb",
+    "rijkswet": "rijkswet",
+    "verdrag": "verdrag",
+    "beleidsregel": "beleidsregel",
+    "circulaire": "circulaire",
+    "zbo": "zbo",
+    "pbo": "pbo",
+}
+
+
+def run_pipeline(
+    output_dir: Path,
+    xml_cache_dir: Path | None = None,
+    types: list[str] | None = None,
+    limit: int | None = None,
+    delay: float = 0.3,
+) -> dict[str, int]:
+    """Draai de volledige BWB pipeline.
+
+    Args:
+        output_dir: Waar de Markdown bestanden geschreven worden.
+        xml_cache_dir: Cache directory voor gedownloade XML (default: output_dir/.xml-cache).
+        types: Regelingtypen om te verwerken (default: alle).
+        limit: Maximum aantal regelingen (None = alles).
+        delay: Vertraging tussen downloads (sec).
+
+    Returns:
+        Dict met statistieken (success, failed, skipped).
+    """
+    xml_cache_dir = xml_cache_dir or output_dir / ".xml-cache"
+    xml_cache_dir.mkdir(parents=True, exist_ok=True)
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    stats = {"catalogus": 0, "downloaded": 0, "parsed": 0, "failed": 0, "skipped": 0}
+
+    # Stap 1: SRU catalogus ophalen
+    logger.info("Stap 1: SRU catalogus ophalen (types=%s, limit=%s)", types, limit)
+    records = fetch_catalogue(types=types, max_records=limit)
+    stats["catalogus"] = len(records)
+    logger.info("Catalogus: %d regelingen gevonden", len(records))
+
+    # Dedupliceer op BWB-ID (SRU kan meerdere toestanden per regeling geven)
+    # Neem de laatste (meest recente) per BWB-ID
+    seen: dict[str, SRURecord] = {}
+    for r in records:
+        seen[r.bwb_id] = r  # Laatste wint
+    unique_records = list(seen.values())
+    logger.info("Na deduplicatie: %d unieke regelingen", len(unique_records))
+
+    # Stap 2+3: Download en parse
+    for i, record in enumerate(unique_records):
+        if i > 0 and i % 100 == 0:
+            logger.info("Voortgang: %d/%d (%d parsed, %d failed)",
+                        i, len(unique_records), stats["parsed"], stats["failed"])
+
+        # Bepaal output pad: output_dir/{type}/{slug}/{BWB_ID}/README.md
+        type_dir = TYPE_TO_DIR.get(record.type, "overig")
+        slug = _slugify(record.titel) if record.titel else record.bwb_id.lower()
+        regeling_dir = output_dir / type_dir / slug / record.bwb_id
+        md_path = regeling_dir / "README.md"
+
+        if md_path.exists():
+            stats["skipped"] += 1
+            continue
+
+        # Download
+        xml_path = download_xml(record, xml_cache_dir)
+        if xml_path is None:
+            stats["failed"] += 1
+            continue
+        stats["downloaded"] += 1
+
+        # Parse
+        try:
+            result = parse_bwb_xml(str(xml_path))
+            regeling_dir.mkdir(parents=True, exist_ok=True)
+            md_path.write_text(result.markdown, encoding="utf-8")
+            stats["parsed"] += 1
+        except Exception as e:
+            logger.warning("Parse failed for %s: %s", record.bwb_id, e)
+            stats["failed"] += 1
+
+        if delay > 0:
+            time.sleep(delay)
+
+    logger.info(
+        "Pipeline klaar: %d parsed, %d failed, %d skipped van %d",
+        stats["parsed"], stats["failed"], stats["skipped"], stats["catalogus"],
+    )
+    return stats
+
+
+def _slugify(text: str) -> str:
+    """Maak een URL-veilige slug van een titel."""
+    import re
+    slug = text.lower().strip()
+    slug = re.sub(r"[^\w\s-]", "", slug)
+    slug = re.sub(r"[\s_]+", "-", slug)
+    slug = re.sub(r"-+", "-", slug)
+    return slug[:80].strip("-")
+
+
+if __name__ == "__main__":
+    import argparse
+
+    logging.basicConfig(
+        level=logging.INFO,
+        format="%(asctime)s %(levelname)s %(message)s",
+        datefmt="%H:%M:%S",
+    )
+
+    parser = argparse.ArgumentParser(description="WetGit BWB pipeline")
+    parser.add_argument("--output", type=Path, required=True, help="Output directory")
+    parser.add_argument("--xml-cache", type=Path, help="XML cache directory")
+    parser.add_argument("--type", action="append", dest="types", help="Regelingtype (herhaalbaar)")
+    parser.add_argument("--limit", type=int, help="Maximum aantal regelingen")
+    parser.add_argument("--delay", type=float, default=0.3, help="Delay tussen downloads (sec)")
+    args = parser.parse_args()
+
+    stats = run_pipeline(
+        output_dir=args.output,
+        xml_cache_dir=args.xml_cache,
+        types=args.types,
+        limit=args.limit,
+        delay=args.delay,
+    )
+    print(json.dumps(stats, indent=2))
diff --git a/src/wetgit/pipeline/sru_client.py b/src/wetgit/pipeline/sru_client.py
new file mode 100644
index 0000000..08e53c4
--- /dev/null
+++ b/src/wetgit/pipeline/sru_client.py
@@ -0,0 +1,186 @@
+"""SRU client voor het Basiswettenbestand.
+
+Crawlt de SRU-zoekservice van overheid.nl om BWB regelingen te vinden
+en download-URLs voor de XML-toestanden op te halen.
+"""
+
+from __future__ import annotations
+
+import logging
+import time
+from dataclasses import dataclass, field
+
+import httpx
+from lxml import etree
+
+logger = logging.getLogger(__name__)
+
+SRU_BASE = "https://zoekservice.overheid.nl/sru/Search"
+SRU_NS = {
+    "srw": "http://www.loc.gov/zing/srw/",
+    "bwb": "http://standaarden.overheid.nl/bwb/terms/",
+    "dcterms": "http://purl.org/dc/terms/",
+    "overheid": "http://standaarden.overheid.nl/owms/terms/",
+}
+
+# Alle regelingtypen in het BWB
+BWB_TYPES = [
+    "wet", "AMvB", "ministeriele-regeling", "KB", "rijkswet",
+    "verdrag", "beleidsregel", "circulaire", "zbo", "pbo",
+]
+
+
+@dataclass(frozen=True)
+class SRURecord:
+    """Eén regeling uit de SRU-response."""
+
+    bwb_id: str
+    titel: str
+    type: str
+    xml_url: str
+    wti_url: str
+    datum_geldig_van: str | None = None
+    datum_geldig_tot: str | None = None
+    ministerie: str | None = None
+
+
+def fetch_catalogue(
+    types: list[str] | None = None,
+    max_records: int | None = None,
+    batch_size: int = 100,
+    delay: float = 0.5,
+) -> list[SRURecord]:
+    """Haal de volledige BWB catalogus op via SRU.
+
+    Args:
+        types: Lijst van regelingtypen om op te halen (default: alle).
+        max_records: Maximum aantal records (None = alles).
+        batch_size: Aantal records per SRU-request (max 100).
+        delay: Vertraging tussen requests (sec) om de server niet te overbelasten.
+
+    Returns:
+        Lijst van SRURecords met metadata en download-URLs.
+    """
+    types = types or BWB_TYPES
+    all_records: list[SRURecord] = []
+
+    for type_name in types:
+        query = f"dcterms.type=={type_name}"
+        start = 1
+
+        while True:
+            records, total = _fetch_page(query, start, batch_size)
+            all_records.extend(records)
+            logger.info(
+                "SRU %s: %d-%d / %d (totaal: %d)",
+                type_name, start, start + len(records) - 1, total, len(all_records),
+            )
+
+            if max_records and len(all_records) >= max_records:
+                return all_records[:max_records]
+
+            start += len(records)
+            if start > total or not records:
+                break
+
+            time.sleep(delay)
+
+    return all_records
+
+
+def fetch_latest_toestand(bwb_id: str) -> SRURecord | None:
+    """Haal de meest recente toestand op voor één BWB-ID.
+
+    Args:
+        bwb_id: Het BWB identificatienummer (bijv. BWBR0001840).
+
+    Returns:
+        SRURecord of None als niet gevonden.
+    """
+    records, _ = _fetch_page(f"dcterms.identifier={bwb_id}", 1, 100)
+    if not records:
+        return None
+    # Neem de laatste (meest recente toestand)
+    return records[-1]
+
+
+def _fetch_page(
+    query: str, start_record: int, maximum_records: int,
+) -> tuple[list[SRURecord], int]:
+    """Haal één pagina SRU-resultaten op."""
+    params = {
+        "operation": "searchRetrieve",
+        "version": "1.2",
+        "x-connection": "BWB",
+        "query": query,
+        "startRecord": str(start_record),
+        "maximumRecords": str(maximum_records),
+    }
+
+    resp = httpx.get(SRU_BASE, params=params, timeout=30)
+    resp.raise_for_status()
+    tree = etree.fromstring(resp.content)
+
+    # Totaal aantal records
+    total_el = tree.find(".//srw:numberOfRecords", SRU_NS)
+    total = int(total_el.text) if total_el is not None and total_el.text else 0
+
+    records: list[SRURecord] = []
+    for record in tree.findall(".//srw:record", SRU_NS):
+        parsed = _parse_record(record)
+        if parsed:
+            records.append(parsed)
+
+    return records, total
+
+
+def _parse_record(record: etree._Element) -> SRURecord | None:
+    """Parse één SRU record naar een SRURecord."""
+    # Zoek in originalData en enrichedData
+    bwb_id = ""
+    titel = ""
+    type_ = ""
+    xml_url = ""
+    wti_url = ""
+    datum_van = None
+    datum_tot = None
+    ministerie = None
+
+    for elem in record.iter():
+        tag = elem.tag
+        text = (elem.text or "").strip()
+
+        if not text:
+            continue
+
+        if tag.endswith("}identifier") and "BWB" in text:
+            bwb_id = text
+        elif tag.endswith("}title"):
+            titel = text
+        elif tag.endswith("}type") and text in BWB_TYPES:
+            type_ = text
+        elif tag.endswith("}authority") or tag.endswith("}creator"):
+            if not ministerie:
+                ministerie = text
+        elif "locatie_toestand" in tag:
+            xml_url = text
+        elif "locatie_wti" in tag:
+            wti_url = text
+        elif "geldigheidsperiode_startdatum" in tag:
+            datum_van = text
+        elif "geldigheidsperiode_einddatum" in tag:
+            datum_tot = text
+
+    if not bwb_id or not xml_url:
+        return None
+
+    return SRURecord(
+        bwb_id=bwb_id,
+        titel=titel,
+        type=type_,
+        xml_url=xml_url,
+        wti_url=wti_url,
+        datum_geldig_van=datum_van,
+        datum_geldig_tot=datum_tot,
+        ministerie=ministerie,
+    )