diff --git a/tools/extract-metadata.py b/tools/extract-metadata.py
new file mode 100755
index 00000000..3656a7d3
--- /dev/null
+++ b/tools/extract-metadata.py
@@ -0,0 +1,211 @@
+#!/usr/bin/env python3
+import pathlib
+import xml.dom.minidom
+
+import xml.etree.ElementTree as etree
+
+
+DESCRIPTION = """\
+Extract a list of XEPs with metadata from the xeps repository."""
+
+EPILOG = """"""
+
+
+def open_xml(f):
+    return xml.dom.minidom.parse(f)
+
+
+def find_child(elem, child_tag):
+    for child in elem.childNodes:
+        if hasattr(child, "tagName") and child.tagName == child_tag:
+            return child
+    return None
+
+
+def find_header(document):
+    header = find_child(document.documentElement, "header")
+    if header is None:
+        raise ValueError("cannot find <header/>")
+    return header
+
+
+def get_text(elem):
+    return "".join(
+        child.nodeValue
+        for child in elem.childNodes
+        if isinstance(child, (xml.dom.minidom.Text,
+                              xml.dom.minidom.CDATASection))
+    )
+
+
+def children(elem):
+    return [
+        child for child in elem.childNodes
+        if isinstance(child, (xml.dom.minidom.Element))
+    ]
+
+
+def extract_xep_metadata(document):
+    header = find_header(document)
+
+    latest_revision = find_child(header, "revision")
+    if latest_revision is not None:
+        last_revision_version = get_text(find_child(latest_revision, "version"))
+        last_revision_date = get_text(find_child(latest_revision, "date"))
+        remark_el = find_child(latest_revision, "remark")
+        last_revision_remark = None
+        if remark_el is not None:
+            remark_children = children(remark_el)
+            if len(remark_children) == 1 and remark_children[0].tagName == "p":
+                last_revision_remark = get_text(remark_children[0])
+
+        if last_revision_remark is not None:
+            initials_el = find_child(latest_revision, "initials")
+            last_revision_initials = initials_el and get_text(initials_el)
+        else:
+            last_revision_initials = None
+    else:
+        last_revision_version = None
+        last_revision_date = None
+        last_revision_remark = None
+        last_revision_initials = None
+
+    status = get_text(find_child(header, "status"))
+    type_ = get_text(find_child(header, "type"))
+    abstract = " ".join(get_text(find_child(header, "abstract")).split())
+    sig_el = find_child(header, "sig")
+    if sig_el is None:
+        sig = None
+    else:
+        sig = get_text(sig_el)
+    shortname = get_text(find_child(header, "shortname"))
+    if shortname.replace("-", " ").replace("_", " ").lower() in [
+            "not yet assigned", "n/a", "none", "to be assigned",
+            "to be issued"]:
+        shortname = None
+    title = get_text(find_child(header, "title"))
+
+    approver_el = find_child(header, "approver")
+    if approver_el is not None:
+        approver = get_text(approver_el)
+    else:
+        approver = "Board" if type_ == "Procedural" else "Council"
+
+    return {
+        "last_revision": {
+            "version": last_revision_version,
+            "date": last_revision_date,
+            "initials": last_revision_initials,
+            "remark": last_revision_remark,
+        },
+        "status": status,
+        "type": type_,
+        "sig": sig,
+        "abstract": abstract,
+        "shortname": shortname,
+        "title": title,
+        "approver": approver,
+    }
+
+
+def text_element(tag, text):
+    el = etree.Element(tag)
+    el.text = text
+    return el
+
+
+def make_metadata_element(number, metadata, accepted, *, protoname=None):
+    result = etree.Element("xep")
+    result.append(text_element("number", number))
+    result.append(text_element("title", metadata["title"]))
+    result.append(text_element("abstract", metadata["abstract"]))
+    result.append(text_element("type", metadata["type"]))
+    result.append(text_element("status", metadata["status"]))
+    result.append(text_element("approver", metadata["approver"]))
+
+    if metadata["shortname"] is not None:
+        result.append(text_element("shortname", metadata["shortname"]))
+
+    if metadata["last_revision"]["version"] is not None:
+        last_revision = metadata["last_revision"]
+        revision_el = etree.Element("last-revision")
+        revision_el.append(text_element("date", last_revision["date"]))
+        revision_el.append(text_element("version", last_revision["version"]))
+        if last_revision["initials"]:
+            revision_el.append(text_element("initials",
+                                            last_revision["initials"]))
+        if last_revision["remark"]:
+            revision_el.append(text_element("remark",
+                                            last_revision["remark"]))
+        result.append(revision_el)
+
+    if metadata["sig"] is not None:
+        result.append(
+            text_element("sig", metadata["sig"])
+        )
+
+    if accepted:
+        result.set("accepted", "true")
+    else:
+        result.set("accepted", "false")
+
+    if protoname is not None:
+        result.append(text_element("proto-name", protoname))
+
+    return result
+
+
+def main():
+    import argparse
+    import sys
+
+    parser = argparse.ArgumentParser(
+        description=DESCRIPTION,
+        epilog=EPILOG,
+    )
+    parser.add_argument(
+        "xepdir",
+        nargs="?",
+        type=pathlib.Path,
+        default=pathlib.Path.cwd(),
+        help="Directory where the XEP XMLs are. Defaults to current directory."
+    )
+
+    args = parser.parse_args()
+
+    tree = etree.Element("xep-infos")
+
+    for xepfile in args.xepdir.glob("xep-*.xml"):
+        number = xepfile.name.split("-", 1)[1].split(".", 1)[0]
+        try:
+            number = str(int(number))
+        except ValueError:
+            continue
+
+        with xepfile.open("rb") as f:
+            parsed = open_xml(f)
+
+        tree.append(make_metadata_element(
+            number,
+            extract_xep_metadata(parsed),
+            True,
+        ))
+
+    for xepfile in (args.xepdir / "inbox").glob("*.xml"):
+        protoname = xepfile.name.rsplit(".", 1)[0]
+
+        with xepfile.open("rb") as f:
+            parsed = open_xml(f)
+
+        tree.append(make_metadata_element(
+            "xxxx",
+            extract_xep_metadata(parsed),
+            False,
+            protoname=protoname
+        ))
+
+    sys.stdout.buffer.raw.write(etree.tostring(tree))
+
+
+if __name__ == "__main__":
+    main()