---
id: "CVE-2026-126393"
page: "https://infrastack.ai/live/CVE-2026-126393"
kind: "vulnerability"
aliases: ["CVE-2026-126393"]
title: "SiYuan: skill-install download lets attackers reach internal services"
severity: "medium"
cvss: 6.1
epss: null
exploited: false
status: "open"
category: "package"
products: []
ecosystems: []
tags: ["mcp", "package", "self-hosted", "siyuan"]
not_affected_if: ["mcp-disabled", "siyuan-3.8.10-or-later"]
published_at: "2026-11-13T10:17:43+00:00"
first_seen_at: "2026-11-13T10:47:18+00:00"
updated_at: "2026-11-13T10:56:27+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-126393", "first_seen_at": "2026-11-13T10:47:18+00:00"}]
exploit_sources: []
---

# SiYuan: skill-install download lets attackers reach internal services

Link for people: https://infrastack.ai/live/CVE-2026-126393 (this .md is for agents)

SiYuan's skill-install download path checks the hostname only when parsing the URL, so redirects or DNS rebinding can send requests to loopback addresses. An attacker who can submit a skill URL through the MCP skill install action can reach internal services and cloud metadata endpoints.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Check the installed SiYuan version; affected builds are before `3.8.10`
- Confirm whether the MCP skill install action is reachable from untrusted users

## Fix

- SiYuan: upgrade to `3.8.10` or later

## Timeline

- 2026-11-13T10:47:18+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-126393

## Documents

- [advisory] Vendor advisory: github.com https://github.com/siyuan-note/siyuan/security/advisories/GHSA-6pg3-pqxw-jm82
- [advisory] Third-party advisory: www.vulncheck.com https://www.vulncheck.com/advisories/siyuan-before-3.8.10-ssrf-via-skill-install-download-redirect-and-dns-rebinding

## From the sources

SiYuan before 3.8.10 contains a server-side request forgery vulnerability in the skill-install download path because fetchBytes performs only a parse-time hostname check instead of using ssrfSafeClient. Attackers can supply a skill URL via the MCP skill install action that redirects or DNS-rebinds to loopback addresses, reaching internal services and metadata endpoints.

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_