diff --git a/.gitignore b/.gitignore
index c601a44..ad5ef82 100644
--- a/.gitignore
+++ b/.gitignore
@@ -128,3 +128,6 @@ dmypy.json
 # Pyre type checker
 .pyre/
 conda_forge_metadata/_version.py
+
+# Downloaded stuff
+.repodata_cache/
diff --git a/conda_forge_metadata/repodata.py b/conda_forge_metadata/repodata.py
new file mode 100644
index 0000000..2206515
--- /dev/null
+++ b/conda_forge_metadata/repodata.py
@@ -0,0 +1,143 @@
+"""
+Utilities to deal with repodata
+"""
+
+import bz2
+import json
+import os
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from functools import lru_cache
+from itertools import product
+from logging import getLogger
+from pathlib import Path
+from typing import Any, Dict, Generator, Iterable, List, Tuple, Union
+from urllib.request import urlretrieve
+
+import bs4
+import requests
+
+logger = getLogger(__name__)
+
+SUBDIRS = (
+    "linux-64",
+    "linux-aarch64",
+    "linux-ppc64le",
+    "osx-64",
+    "osx-arm64",
+    "win-64",
+    "win-arm64",
+    "noarch",
+)
+CACHE_DIR = Path(".repodata_cache")
+
+
+@lru_cache
+def all_labels(use_remote_cache: bool = False) -> List[str]:
+    if use_remote_cache:
+        r = requests.get(
+            "https://raw.githubusercontent.com/conda-forge/"
+            "by-the-numbers/main/data/labels.json"
+        )
+        r.raise_for_status()
+        return r.json()
+
+    if token := os.environ.get("BINSTAR_TOKEN"):
+        label_info = requests.get(
+            "https://api.anaconda.org/channels/conda-forge",
+            headers={"Authorization": f"token {token}"},
+        ).json()
+
+        return sorted(label for label in label_info if "/" not in label)
+
+    logger.info("No token detected. Fetching labels from anaconda.org HTML. Slow...")
+    r = requests.get("https://anaconda.org/conda-forge/repo")
+    r.raise_for_status()
+    html = r.text
+    soup = bs4.BeautifulSoup(html, "html.parser")
+    labels = []
+    len_prefix = len("/conda-forge/repo?label=")
+    for element in soup.select("ul#Label > li > a"):
+        href = element.get("href")
+        if not href:
+            continue
+        label = href[len_prefix:]
+        if label and label not in ("all", "empty") and "/" not in label:
+            labels.append(label)
+    return sorted(labels)
+
+
+def fetch_repodata(
+    subdirs: Iterable[str] = SUBDIRS,
+    force_download: bool = False,
+    cache_dir: Union[str, Path] = CACHE_DIR,
+    label: str = "main",
+) -> List[Path]:
+    assert all(subdir in SUBDIRS for subdir in subdirs)
+    paths = []
+    for subdir in subdirs:
+        if label == "main":
+            repodata = f"https://conda.anaconda.org/conda-forge/{subdir}/repodata.json"
+        else:
+            repodata = (
+                "https://conda.anaconda.org/conda-forge/"
+                f"label/{label}/{subdir}/repodata.json"
+            )
+        local_fn = Path(cache_dir, f"{subdir}.{label}.json")
+        local_fn_bz2 = Path(str(local_fn) + ".bz2")
+        paths.append(local_fn)
+        if force_download or not local_fn.exists():
+            logger.info(f"Downloading {repodata} to {local_fn}")
+            local_fn.parent.mkdir(parents=True, exist_ok=True)
+            # Download the file
+            urlretrieve(f"{repodata}.bz2", local_fn_bz2)
+            with open(local_fn_bz2, "rb") as compressed, open(local_fn, "wb") as f:
+                f.write(bz2.decompress(compressed.read()))
+            local_fn_bz2.unlink()
+    return paths
+
+
+def list_artifacts(
+    repodata_jsons: Iterable[Union[str, Path]],
+    include_broken: bool = True,
+) -> Generator[str, None, None]:
+    for repodata in sorted(repodata_jsons):
+        repodata = Path(repodata)
+        subdir = repodata.stem.split(".")[0]
+        assert (
+            subdir in SUBDIRS
+        ), "Invalid repodata file name. Must be '<subdir>.<label>.json'."
+        data = json.loads(repodata.read_text())
+        keys = ["packages", "packages.conda"]
+        if include_broken:
+            keys.append("removed")
+        for key in keys:
+            for pkg in data.get(key, ()):
+                yield f"{subdir}/{pkg}"
+
+
+def repodata(subdir: str) -> Dict[str, Any]:
+    assert subdir in SUBDIRS
+    path = fetch_repodata(subdirs=(subdir,))[0]
+    return json.loads(path.read_text())
+
+
+def n_artifacts(labels: Iterable[str] = ("main",)) -> Tuple[int, int]:
+    """
+    To get _all_ artifacts ever published, use `n_artifacts(all_labels())`.
+
+    Returns number of artifacts and number of unique package names.
+    """
+    seen_artifacts, seen_package_names = set(), set()
+    futures = []
+    with ThreadPoolExecutor(max_workers=10) as executor:
+        for label, subdir in product(labels, SUBDIRS):
+            future = executor.submit(fetch_repodata, (subdir,), False, CACHE_DIR, label)
+            futures.append(future)
+        for future in as_completed(futures):
+            repodatas = future.result()
+            artifacts = list_artifacts(repodatas, include_broken=True)
+            for artifact in artifacts:
+                seen_artifacts.add(artifact)
+                seen_package_names.add(artifact.split("/")[-1].rsplit("-", 2)[0])
+
+    return len(seen_artifacts), len(seen_package_names)
diff --git a/pyproject.toml b/pyproject.toml
index afe0ad2..f037748 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -17,7 +17,8 @@ dependencies = [
     "deprecated",
     "requests",
     "ruamel.yaml",
-    "typing-extensions"
+    "typing-extensions",
+    "beautifulsoup4"
 ]
 license = {file = "LICENSE"}
 readme = "README.md"
diff --git a/requirements-dev.txt b/requirements-dev.txt
index 71c9853..db118bb 100644
--- a/requirements-dev.txt
+++ b/requirements-dev.txt
@@ -1,7 +1,7 @@
 flake8
 flaky
 pip
-pytest
+pytest <8.1.0a0  # flaky does not support pytest >=8.1
 python-build
 setuptools>=45
 setuptools_scm>=7
diff --git a/requirements.txt b/requirements.txt
index 2ab92ec..297d940 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,3 +1,4 @@
+beautifulsoup4
 conda-oci-mirror
 deprecated
 requests
diff --git a/tests/test_repodata.py b/tests/test_repodata.py
new file mode 100644
index 0000000..83b9cae
--- /dev/null
+++ b/tests/test_repodata.py
@@ -0,0 +1,9 @@
+from conda_forge_metadata import repodata
+
+
+def test_labels_anaconda_org(monkeypatch):  # type: ignore
+    monkeypatch.delenv("BINSTAR_TOKEN", raising=False)
+    labels = repodata.all_labels()
+    assert len(labels) >= 20
+    assert "main" in labels
+    assert "broken" in labels