diff --git a/scrape_repos.py b/scrape_repos.py
new file mode 100644
index 0000000..0c17ab5
--- /dev/null
+++ b/scrape_repos.py
@@ -0,0 +1,84 @@
+"""
+GitHub API repository scraper
+Simple as possible, but not production ready.
+
+"""
+
+import json
+import os
+import time
+import tqdm
+
+from requests.exceptions import Timeout
+from github import Github, Repository, RateLimitExceededException, GithubException
+
+BASE_ID = 3544490  # id of python-github package, because we don't need first repositories
+OUT_PATH = "repos_scraped.jsl"
+TIMEOUT_SLEEP = 5 * 60
+HEADER_SIZE_TO_READ = 3 * 1024
+
+g = Github(os.getenv("GITHUB_KEY"), per_page=100)
+
+
+def write_repo(f, r: Repository):
+    try:
+        data_to_write = (
+            json.dumps(
+                {
+                    "id": r.id,
+                    "name": r.full_name,
+                    "fork": r.fork,
+                    "size": r.size,
+                    "default_branch": r.default_branch,
+                    "stargazers_count": r.stargazers_count,
+                    "updated_at": int(r.updated_at.timestamp()),
+                    "created_at": int(r.created_at.timestamp()),
+                    "private": r.private,
+                    "archived": r.archived,
+                },
+                ensure_ascii=False,
+            )
+            + "\n"
+        )
+    except GithubException:
+        print("error with", r)
+        time.sleep(2)
+        return
+    f.write(data_to_write.encode("utf-8"))
+
+
+def get_last_id(path: str) -> int:
+    if not os.path.exists(path):
+        print("No base file, return base value", BASE_ID)
+        return BASE_ID
+    total_size = os.path.getsize(path)
+    with open(path, "rb") as f:
+        f.seek(max(0, total_size - HEADER_SIZE_TO_READ))
+        data = f.read()
+    if not data:
+        return BASE_ID
+    last_item = json.loads(data.decode("utf-8").splitlines()[-1])
+    return last_item["id"]
+
+
+def main():
+    path = "repos_new.jsl"
+    f = open(path, "ab")
+    while True:
+        last_item_id = get_last_id(path)
+        try:
+            for r in tqdm.tqdm(g.get_repos(since=last_item_id)):
+                write_repo(f, r)
+        except RateLimitExceededException:
+            print("waiting after", get_last_id(path))
+            time.sleep(30 * 60)
+
+
+if __name__ == "__main__":
+    print(get_last_id("repos_new.jsl"))
+    while True:
+        try:
+            main()
+        except Timeout:
+            print("timeout")
+            time.sleep(TIMEOUT_SLEEP)