storyblok-backup: Stored path traversal via unsanitized Storyblok component names in backup outputGHSA-vcx7-pfrh-mcfg
MEDIUMFix: webflorist/storyblok-backup@f4e4bb1GHSA-vcx7-pfrh-mcfg is a medium-severity (CVSS 5.7) Path Traversal vulnerability in storyblok-backup. A fix is available for storyblok-backup — see the affected versions and patch details below.
EPSS Exploitation Probability
Probability of exploitation in the next 30 days, from FIRST.org EPSS.
How urgent is this, really
Where this sits among everything scored
Real-World Exposure
storyblok-backupReal-time download stats are indexed for npm and PyPI packages. This vulnerability affects npm packages — download data is not available via public APIs for these ecosystems.
Description
Summary
storyblok-backup writes files using unsanitized names for Storyblok components, which are content type or schema definitions. This allows a component name containing path traversal sequences to cause backup files to be written outside the intended backup directory. Anyone with component management access in the source Storyblok space, not server access, can plant a malicious component name that causes arbitrary file writes on the machine of anyone who later runs a backup against that space.
Details
The writeJson function, which takes a folder, a file name, and content, builds the output file path by directly concatenating the backup directory, the folder, and the file name as strings, with no path sanitization or containment check.
For the components resource type specifically, the file name used is the component's own name field, which is free text and editable by anyone with component management access in the Storyblok CMS.
Other resource types are not affected by this same issue. Stories and datasources use their numeric IDs as filenames, and assets have their filename derived by splitting on the slash character and taking the last segment, which strips path separators. This issue is scoped specifically to components.
Since the component name is stored in Storyblok and only read back later when the backup tool runs, this is a stored, second order vulnerability. The malicious value is planted once, then triggers on every subsequent backup run performed by anyone using this tool against that Storyblok space.
PoC
Step 1 — plant the malicious component name via Storyblok's Management API:
curl -X POST "https://mapi.storyblok.com/v1/spaces/<SPACE_ID>/components/"
-H "Authorization: <MANAGEMENT_API_TOKEN>"
-H "Content-Type: application/json"
-d '{
"component": {
"name": "../../../../tmp/evil-storyblok-backup",
"schema": {}
}
}'
Any Storyblok user with component-management permission in the space (not server access to wherever backups run) can do this through the Storyblok UI just as easily, by naming a new component with this value in the component name field.
Step 2 — run storyblok-backup against this space, exactly as a normal user would:
npx storyblok-backup
--space <SPACE_ID>
--token <MANAGEMENT_API_TOKEN>
--path ./backup
--types "components"
Step 3 — observe the actual file write location:
With the malicious component name above, the tool's writeJson() function builds the output path as:
backupDir + "/components" + "/../../../../tmp/evil-storyblok-backup" + ".json"
which resolves outside ./backup entirely. Confirm with:
ls -la /tmp/evil-storyblok-backup.json
The file exists there, outside the --path ./backup directory that was configured, containing the full JSON definition of the malicious component.
Step 4 — for comparison, confirm assets and stories are NOT affected:
The same defaultFetch() pattern is used for stories and datasources (fileField: numeric id) and for assets (fileField derived via .split('/').at(-1), which strips any path separators from the value before it reaches writeJson()). Only the components resource type passes a raw, unsanitized user-editable string directly into the file path.
Impact
Anyone with the ability to create or rename a component in the source Storyblok space, which is a content management privilege and not server access, can cause the backup tool, when run by someone else, to write files outside the intended backup directory on whatever machine runs the backup. Depending on the path chosen and the permissions of the process running the backup, this could overwrite existing files at attacker chosen locations on the host filesystem.
Suggested fix: sanitize the component name before using it as a filename, for example by applying the same last-path-segment treatment already used for assets, or by resolving the final path and verifying it stays within the intended backup directory before writing.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 📦npm | storyblok-backup | all versions | 0.7.1npm install storyblok-backup@0.7.1 |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for storyblok-backup, including transitive dependencies — a direct dependency you never call can still pull in a vulnerable version.
Fix
Update storyblok-backup to 0.7.1 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms GHSA-vcx7-pfrh-mcfg is resolved across your whole dependency graph.
Workarounds
Resolve every user-supplied path to its canonical form and reject anything that escapes the intended directory, and run the component under an account that has no read or write access outside the directory it legitimately serves.
Frequently Asked Questions
Is GHSA-vcx7-pfrh-mcfg in your dependencies?
Find it across npm, including transitive dependencies.