GHSA-6JM8-X3G6-R33J
Vulnerability from github – Published: 2026-01-08 21:01 – Updated: 2026-01-08 21:37LFS Lock Force-Delete Authorization Bypass
Summary
An authorization bypass in the LFS lock deletion endpoint allows any authenticated user with repository write access to delete locks owned by other users by setting the force flag. The vulnerable code path processes force deletions before retrieving user context, bypassing ownership validation entirely.
Severity
- CWE-863: Incorrect Authorization
- CVSS 3.1: 5.4 (Medium) —
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L
Affected Code
File: pkg/web/git_lfs.go
Function: serviceLfsLocksDelete (lines 831–945)
Endpoint: POST /<repo>.git/info/lfs/locks/:lockID/unlock
The control flow processes req.Force at line 905 before retrieving user context at line 919:
// Line 905-916: Force delete executes immediately without authorization
if req.Force {
if err := datastore.DeleteLFSLock(ctx, dbx, repo.ID(), lockID); err != nil {
// ...
}
renderJSON(w, http.StatusOK, l)
return // Returns here, never reaching user validation
}
// Line 919: User context retrieved after force path has exited
user := proto.UserFromContext(ctx)
Proof of Concept
Setup: Two users with write access to the same repository—User A (lock owner) and User B (attacker).
-
User A creates a lock:
bash curl -X POST http://localhost:23232/repo.git/info/lfs/locks \ -H "Authorization: Basic <user_a_token>" \ -H "Content-Type: application/vnd.git-lfs+json" \ -d '{"path": "protected-file.bin"}' -
User B deletes User A's lock using force flag:
bash curl -X POST http://localhost:23232/repo.git/info/lfs/locks/1/unlock \ -H "Authorization: Basic <user_b_token>" \ -H "Content-Type: application/vnd.git-lfs+json" \ -d '{"force": true}' -
Result: Lock deleted successfully with
200 OK. Expected:403 Forbidden.
Suggested Fix
Retrieve user context and validate authorization before processing the force flag:
user := proto.UserFromContext(ctx)
if user == nil {
renderJSON(w, http.StatusUnauthorized, lfs.ErrorResponse{
Message: "unauthorized",
})
return
}
if req.Force {
if !user.IsAdmin() {
renderJSON(w, http.StatusForbidden, lfs.ErrorResponse{
Message: "admin access required for force delete",
})
return
}
if err := datastore.DeleteLFSLock(ctx, dbx, repo.ID(), lockID); err != nil {
// ...
}
renderJSON(w, http.StatusOK, l)
return
}
Impact
Affected Deployments: Soft Serve instances with LFS enabled and repositories with multiple collaborators.
Exploitation Requirements: - Authenticated session - Write access to target repository
Consequences: - Unauthorized deletion of other users' locks - Bypass of LFS file coordination mechanisms - Potential workflow disruption in collaborative environments
Limitations: Does not grant file access, escalate repository permissions, or affect repositories where the attacker lacks write access.
{
"affected": [
{
"package": {
"ecosystem": "Go",
"name": "github.com/charmbracelet/soft-serve"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.11.2"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-22253"
],
"database_specific": {
"cwe_ids": [
"CWE-863"
],
"github_reviewed": true,
"github_reviewed_at": "2026-01-08T21:01:54Z",
"nvd_published_at": "2026-01-08T19:15:59Z",
"severity": "MODERATE"
},
"details": "## LFS Lock Force-Delete Authorization Bypass\n\n### Summary\n\nAn authorization bypass in the LFS lock deletion endpoint allows any authenticated user with repository write access to delete locks owned by other users by setting the `force` flag. The vulnerable code path processes force deletions before retrieving user context, bypassing ownership validation entirely.\n\n### Severity\n\n- **CWE-863:** Incorrect Authorization\n- **CVSS 3.1:** 5.4 (Medium) \u2014 `CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L`\n\n### Affected Code\n\n**File:** `pkg/web/git_lfs.go`\n**Function:** `serviceLfsLocksDelete` (lines 831\u2013945)\n**Endpoint:** `POST /\u003crepo\u003e.git/info/lfs/locks/:lockID/unlock`\n\nThe control flow processes `req.Force` at line 905 before retrieving user context at line 919:\n\n```go\n// Line 905-916: Force delete executes immediately without authorization\nif req.Force {\n if err := datastore.DeleteLFSLock(ctx, dbx, repo.ID(), lockID); err != nil {\n // ...\n }\n renderJSON(w, http.StatusOK, l)\n return // Returns here, never reaching user validation\n}\n\n// Line 919: User context retrieved after force path has exited\nuser := proto.UserFromContext(ctx)\n```\n\n### Proof of Concept\n\n**Setup:** Two users with write access to the same repository\u2014User A (lock owner) and User B (attacker).\n\n1. **User A creates a lock:**\n ```bash\n curl -X POST http://localhost:23232/repo.git/info/lfs/locks \\\n -H \"Authorization: Basic \u003cuser_a_token\u003e\" \\\n -H \"Content-Type: application/vnd.git-lfs+json\" \\\n -d \u0027{\"path\": \"protected-file.bin\"}\u0027\n ```\n\n2. **User B deletes User A\u0027s lock using force flag:**\n ```bash\n curl -X POST http://localhost:23232/repo.git/info/lfs/locks/1/unlock \\\n -H \"Authorization: Basic \u003cuser_b_token\u003e\" \\\n -H \"Content-Type: application/vnd.git-lfs+json\" \\\n -d \u0027{\"force\": true}\u0027\n ```\n\n3. **Result:** Lock deleted successfully with `200 OK`. Expected: `403 Forbidden`.\n\n### Suggested Fix\n\nRetrieve user context and validate authorization before processing the force flag:\n\n```go\nuser := proto.UserFromContext(ctx)\nif user == nil {\n renderJSON(w, http.StatusUnauthorized, lfs.ErrorResponse{\n Message: \"unauthorized\",\n })\n return\n}\n\nif req.Force {\n if !user.IsAdmin() {\n renderJSON(w, http.StatusForbidden, lfs.ErrorResponse{\n Message: \"admin access required for force delete\",\n })\n return\n }\n if err := datastore.DeleteLFSLock(ctx, dbx, repo.ID(), lockID); err != nil {\n // ...\n }\n renderJSON(w, http.StatusOK, l)\n return\n}\n```\n\n### Impact\n\n**Affected Deployments:** Soft Serve instances with LFS enabled and repositories with multiple collaborators.\n\n**Exploitation Requirements:**\n- Authenticated session\n- Write access to target repository\n\n**Consequences:**\n- Unauthorized deletion of other users\u0027 locks\n- Bypass of LFS file coordination mechanisms\n- Potential workflow disruption in collaborative environments\n\n**Limitations:** Does not grant file access, escalate repository permissions, or affect repositories where the attacker lacks write access.",
"id": "GHSA-6jm8-x3g6-r33j",
"modified": "2026-01-08T21:37:08Z",
"published": "2026-01-08T21:01:54Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/charmbracelet/soft-serve/security/advisories/GHSA-6jm8-x3g6-r33j"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-22253"
},
{
"type": "WEB",
"url": "https://github.com/charmbracelet/soft-serve/commit/000ab5164f0be68cf1ea6b6e7227f11c0e388a42"
},
{
"type": "PACKAGE",
"url": "https://github.com/charmbracelet/soft-serve"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L",
"type": "CVSS_V3"
}
],
"summary": "Soft Serve is missing an authorization check in LFS lock deletion"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.