Skip to main content
Mallory
CriticalExploited in the wildCISA KEVPublic exploitVulnerability

CVE-2025-3248: Unauthenticated Remote Code Execution in Langflow Code Validation

Published Apr 7, 2025 · Updated today
Trending this weekNot trendingOutside this week’s top 100
Mention activity · 30 days14 mentions · peak 3 in a dayToday
CVSS9.8Critical · v3.1
EPSS100.0%100th percentile
CISA KEVYesAdded May 5, 2025
In the wildActivePer CISA KEV
Vendor
Langflow
IdentifiersCVE-2025-3248CWE-94 · Improper Control of Generation of…

CVE-2025-3248 is an unauthenticated Python code-injection vulnerability affecting Langflow versions before 1.3.0 and langflow-base versions before 0.3.0. The code-validation endpoint parses attacker-supplied Python with ast.parse(), compiles function definitions, and executes them using exec() without requiring authentication or adequately isolating execution. Malicious expressions in function decorators or default arguments execute during validation, without the function being called. Crafted HTTP requests therefore allow remote attackers to execute arbitrary code with the Langflow process's privileges. The vulnerability is actively exploited and is listed in CISA's Known Exploited Vulnerabilities catalog.

Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Analyst brief

Impact, mitigation & remediation

What it means and what to do now: patch path, interim mitigations, and the assume-compromise checklist.

Impact

What an attacker gets, and what they’ve been doing with it.

Successful exploitation compromises the confidentiality, integrity, and availability of the Langflow application and resources accessible to its service account. Attackers can read application data, environment variables, stored credentials, and configuration secrets; modify or delete accessible data; and execute additional payloads. Exposed cloud credentials, AI-provider keys, database credentials, or container-management interfaces can enable further compromise, subject to their permissions and network reachability. Observed exploitation includes credential harvesting, Flodrix botnet deployment, cryptocurrency-mining attempts, and ransomware operations.

Mitigation

If you can’t patch tonight, do this now.

Remove direct internet exposure and restrict access to trusted networks, a VPN, or an authenticated access gateway. Enforce authentication before requests reach the code-validation endpoint. Run Langflow with least privilege, avoid mounting container-management sockets, and restrict access to cloud metadata services where possible. Monitor suspicious code-validation requests, unexpected processes, persistence mechanisms, outbound connections, and unusual cloud or AI-service usage. These controls reduce exposure but do not replace upgrading.

Remediation

Patch, then assume compromise.

Upgrade the langflow package to version 1.3.0 or later and langflow-base to version 0.3.0 or later, as applicable. Langflow 1.3.0 adds an active-user authentication dependency to the vulnerable endpoint. Investigate previously exposed instances for compromise, since upgrading does not remove existing attacker access. Rotate potentially exposed provider and cloud credentials before rotating the Langflow secret key, and review application and cloud activity for unauthorized use.

Public exploits

Exploits

27 exploits validated by Mallory's analysis. Hidden: 4 rejected by Mallory's analysis as not a working exploit; 5 not yet analyzed, such as Exploit-DB entries. Showing the 25 most capable, earliest first.

View more in app →
langflow_unauth_rce_cve_2025_3248.rbThis repository contains a single Metasploit module targeting Langflow AI (versions prior to 1.3.0) for unauthenticated remote code execution (RCE), tracked as CVE-2025-3248. The exploit leverages a vulnerability in the /api/v1/validate/code endpoint, allowing an attacker to send a crafted POST request containing arbitrary Python code, which is then executed on the server. The module includes a check method to determine if the target is running a vulnerable version and whether authentication is enabled. The default payload is a Python Meterpreter reverse shell, but any Python code can be executed. The exploit is fully weaponized, allowing for easy payload customization and reliable exploitation. The only file in the repository is a Ruby script compatible with the Metasploit framework, and it interacts with three main HTTP endpoints on the target: /api/v1/validate/code (for exploitation), /api/v1/version (for version detection), and /api/v1/auto_login (to check authentication status).Disclosed Apr 9, 2025 · Framework: metasploit · ruby · networkWeaponizedCVE-2025-3248This repository is an operational exploit and scanner for CVE-2025-3248, a code execution vulnerability in Langflow (<=1.3.0). The main entry point is main.py, which provides a CLI for scanning single or multiple targets (using FOFA for mass discovery). The exploit works by sending a crafted Python payload to the /api/v1/validate/code endpoint, leveraging unsafe code execution to run arbitrary system commands. The tool includes built-in payloads for retrieving system information (uname -a) and reading /etc/passwd, and supports custom payloads via user-supplied Python files. The scanner can output user account and system details if exploitation is successful. The codebase is modular, with separate files for payload generation, FOFA integration, and display utilities. The README provides clear usage instructions, legal disclaimers, and references to the CVE and upstream Langflow project.xuemian168 · Disclosed Apr 10, 2025 · python · networkOperationalCVE-2025-3248This repository contains a working proof-of-concept exploit for CVE-2025-3248, a critical remote code execution vulnerability in Langflow versions prior to 1.3.0. The exploit is implemented in a single Python script (cve-2025-3248.py) and is accompanied by a README.md with usage instructions and background information. The exploit works by sending a crafted POST request to the /api/v1/validate/code endpoint of a vulnerable Langflow instance, injecting a Python function that executes an arbitrary shell command via subprocess. The output of the command is extracted from the error message in the JSON response. The exploit does not require authentication and can be used remotely, making it highly impactful. The repository is well-structured, with clear instructions and a single, focused exploit script. No hardcoded IPs or domains are present, but the targeted endpoint is clearly fingerprintable. The exploit is operational, providing real command execution and output retrieval, but is not part of a larger framework.verylazytech · Disclosed Apr 16, 2025 · python · markdown · networkOperationalCVE-2025-3248This repository contains a functional exploit for CVE-2025-3248, targeting Langflow versions prior to 1.3.0. The exploit leverages a code injection vulnerability in the /api/v1/validate/code API endpoint, allowing a remote, unauthenticated attacker to execute arbitrary Python code on the server. The exploit script (exploit.py) first tests for vulnerability by sending a payload that attempts to execute the 'id' command. If the target is vulnerable, the user is prompted to proceed with exploitation, which involves sending a payload that establishes a reverse shell from the target to the attacker's specified IP and port. The repository consists of a README.md with usage instructions and background, and a single Python exploit script. The exploit is operational, providing a working reverse shell payload, and requires the attacker to set up a listener to receive the shell. The main fingerprintable endpoint is the /api/v1/validate/code API path.Praison001 · Disclosed May 5, 2025 · python · networkOperationalCVE-2025-3248This repository provides a working exploit for CVE-2025-3248, a critical remote code execution vulnerability in Langflow versions 0 through 1.2.0. The exploit is implemented in a single Python script ('exploit'), which targets the /api/v1/validate/code endpoint of a vulnerable Langflow server. The exploit works by sending a specially crafted payload that abuses unsafe code execution in the API, resulting in a reverse shell being established from the target server to the attacker's machine. The script includes logic to detect the correct protocol (HTTP/HTTPS), verify the existence of the vulnerable endpoint, and launch a listener for the reverse shell. The repository also contains a README.md with detailed vulnerability and usage information, and a Docker/setup file to help set up a vulnerable test environment. The exploit is operational and provides full remote code execution if the target is vulnerable and accessible.vigilante-1337 · Disclosed May 13, 2025 · python · bash · networkOperationalCVE-2025-3248This repository contains a functional exploit for CVE-2025-3248, targeting the Langflow API's /api/v1/validate/code endpoint. The main file, CVE-2025-3248.py, is a Python script that allows an attacker to perform remote code execution (RCE) by injecting malicious Python code via a POST request to the vulnerable API endpoint. The script provides an interactive CLI for the user to either execute arbitrary system commands or spawn a reverse shell on the target system. It handles retries if the endpoint is temporarily unavailable and disables SSL verification for convenience. The exploit is operational, providing real payloads for both command execution and reverse shell, and is not just a proof of concept. The README provides basic usage instructions and a disclaimer. No hardcoded IPs or domains are present; the target URL is supplied by the user at runtime.Vip3rLi0n · Disclosed May 27, 2025 · python · networkOperationalRCE-CVE-2025-3248This repository contains an operational exploit for CVE-2025-3248, targeting Langflow version 1.2.0. The main exploit script (exploit.py) is a Python 3 tool that requires valid admin credentials to authenticate to the target Langflow instance. Once authenticated, it abuses the /api/v1/validate/code API endpoint to execute arbitrary system commands or spawn a reverse shell on the target system. The script supports proxying requests (e.g., through Burp Suite) and can be used to either run a single command or establish a reverse shell to an attacker-controlled host and port. The repository also includes a Docker Compose file for deploying a vulnerable Langflow instance locally for testing. The exploit is not part of a framework and is a standalone operational tool. The main attack vector is network-based, requiring access to the target's web interface. Key endpoints include the authentication and code validation APIs. The exploit is well-documented and provides clear usage instructions.tiemio · Disclosed May 31, 2025 · python · yaml · networkOperationalCVE-2025-3248-Langflow-RCEThis repository contains a working exploit for CVE-2025-3248, a remote code execution (RCE) vulnerability in the Langflow application. The exploit is implemented as a single Python script (langflow_rce_unauth.py) and is accompanied by a README.md with usage instructions and vulnerability details. The exploit targets the '/api/v1/validate/code' endpoint of a vulnerable Langflow instance, sending a specially crafted payload that leverages Python's exec() and subprocess modules to execute arbitrary shell commands on the server. No authentication is required, making this a critical unauthenticated RCE. The script provides colorized output and is easy to use, requiring only the target URL and the desired command. The repository is well-structured, with clear documentation and a single, focused exploit script.ynsmroztas · Disclosed Jun 17, 2025 · python · markdown · networkOperationalCVE-2025-3248This repository contains a Python exploit script (CVE-2025-3248.py) and a README.md. The exploit targets Langflow instances vulnerable to CVE-2025-3248, which allows remote, unauthenticated attackers to execute arbitrary system commands via a crafted POST request to the /api/v1/validate/code endpoint. The script supports both single-target and bulk scanning modes, accepts custom commands for execution, and saves results (including command output and scan statistics) to timestamped files. The exploit works by injecting Python code that uses subprocess.check_output to run arbitrary shell commands on the server. The README provides usage instructions and example output. No hardcoded IPs or domains are present; the script is designed to be used against user-supplied targets. The main entry point is CVE-2025-3248.py, which implements all exploit logic.imbas007 · Disclosed Jun 18, 2025 · python · markdown · networkOperationalCVE-2025-3248This repository contains a Python exploit script (CVE-2025-3248.py) targeting a remote code execution (RCE) vulnerability in Langflow applications (CVE-2025-3248). The exploit abuses the /api/v1/validate/code API endpoint, which is accessible without authentication, to execute arbitrary system commands on the target server. The script allows the attacker to specify a single target URL or a list of targets, as well as the command to execute. It supports optional proxying (e.g., through Burp Suite) and provides colorized terminal output for clarity. The README provides usage instructions, affected endpoint details, and search dorks for finding vulnerable instances. The repository is well-structured, with a single exploit script, a README, a license, and a .gitignore. The exploit is operational and returns the output of executed commands, making it a practical tool for attackers or penetration testers.zapstiko · Disclosed Jun 19, 2025 · python · networkOperationalAshwesker-CVE-2025-3248This repository contains a working exploit for CVE-2025-3248, a critical unauthenticated remote code execution (RCE) vulnerability in Langflow (<= v1.3.0). The exploit is implemented in a single Python script (CVE-2025-3248.py) and is accompanied by a detailed README.md. The exploit works by sending a crafted POST request to the vulnerable Langflow endpoint (/api/v1/validate/code), embedding a Python payload that executes arbitrary shell commands via subprocess. The script supports targeting single or multiple URLs, optional proxying, and outputs the result of the executed command. No authentication is required, and the exploit is effective against any exposed, unpatched Langflow instance. The README provides technical details, usage instructions, and search dorks for finding vulnerable targets. The repository is operational and provides a real, working exploit for the specified CVE.Ashwesker · Disclosed Jun 22, 2025 · python · markdown · networkOperationallangflow-rce-exploitThis repository provides a fully operational remote code execution (RCE) exploit for Langflow, targeting CVE-2025-3248. The main exploit script (exploit.py) is a Python 3 tool that interacts with the vulnerable Langflow API endpoint (/api/v1/validate/code) to inject and execute arbitrary Python code on the server. The exploit supports multiple modes: interactive shell, single command execution, reverse shell (with several payload methods), file upload/download (using base64 encoding), and persistence via cron jobs. It features payload obfuscation, header rotation for stealth, and session logging. The repository includes a README.md with detailed usage instructions, requirements.txt for dependencies, and output.txt (empty). The exploit is not part of a framework and is self-contained. The attack vector is network-based, requiring access to the target's HTTP(S) API. The exploit is mature (operational), providing real-world attack capabilities beyond a simple proof of concept.0-d3y · Disclosed Jun 23, 2025 · python · networkOperationalCVE-2025-3248This repository targets CVE-2025-3248, a critical unauthenticated remote code execution (RCE) vulnerability in Langflow AI. It contains two main Python scripts: 1. LangflowCheck.py: A scanner that tests if a given Langflow AI instance is vulnerable by sending a payload to the /api/v1/validate/code endpoint. The payload attempts to read /etc/passwd, and the script checks the response for evidence of successful code execution. It supports scanning single or multiple targets and can output a list of vulnerable hosts. 2. exploit.py: An exploit script that sends a reverse shell payload to the same endpoint. The attacker specifies their own IP and port (lhost/lport), and the script attempts to connect back to the attacker's listener, providing a shell on the target system. The script includes logging and can verify if the attacker's listener is ready before sending the payload. The exploit is operational and provides a working reverse shell if the target is vulnerable. The main attack vector is network-based, targeting the exposed /api/v1/validate/code HTTP endpoint. The repository is well-structured, with clear separation between detection and exploitation scripts, and includes documentation in the README.md. No hardcoded IPs or domains are present; the scripts require user-supplied target and attacker information.r0otk3r · Disclosed Jul 6, 2025 · python · markdown · networkOperationalCVE-2025-3248This repository contains an operational exploit for CVE-2025-3248, targeting Langflow versions prior to 1.3.0. The exploit leverages an unauthenticated code injection vulnerability in the '/api/v1/validate/code' endpoint, which uses Python's exec() function unsafely. The main exploit script, 'cve-2025-3248.py', takes as arguments the target URL, attacker's (C2) IP, and port. It crafts a malicious payload that, when sent to the vulnerable endpoint, causes the server to connect back to the attacker's machine and spawn a reverse shell. The repository includes a README with setup and usage instructions, a requirements.txt for dependencies, and standard project files. The exploit is not part of a framework and is a standalone Python script. The main attack vector is network-based, exploiting a remote API endpoint. The only fingerprintable endpoint is '/api/v1/validate/code'. The exploit is operational, providing a working reverse shell if the target is vulnerable and accessible.wand3rlust · Disclosed Sep 17, 2025 · python · networkOperationalCVE-2025-3248-POCThis repository is a proof-of-concept (PoC) exploit for CVE-2025-3248, a critical remote code execution (RCE) vulnerability in Langflow versions <= 1.3.0. The vulnerability arises from the /api/v1/validate/code endpoint, which executes user-supplied Python code without authentication or input sanitization. The exploit is implemented in Rust (src/main.rs) and allows the user to specify a target URL and a shell command to execute. The exploit crafts a malicious Python payload that is sent via a POST request to the vulnerable endpoint, resulting in arbitrary command execution on the server. The repository includes a Docker Compose file for setting up a vulnerable test environment and a README with detailed usage instructions and vulnerability background. The exploit is operational, providing direct command execution and output retrieval, and targets network-accessible Langflow instances exposing the vulnerable endpoint.b0ySie7e · Disclosed Dec 10, 2025 · rust · yaml · networkOperationalCVE-2025-3248This repository is a small standalone exploit PoC consisting of one Python script and one README. The main file, CVE-2025-3248.py, is a command-line exploit that targets a vulnerable web application's POST endpoint at /api/v1/validate/code. It validates a user-supplied base URL, appends the vulnerable path, and sends a JSON body containing a crafted Python payload. The payload imports os and abuses exec() inside a decorator expression to run os.popen() with an operator-controlled system command. The command output is wrapped in an exception so the result is likely reflected in the HTTP response, which the script prints in either text or JSON format. The exploit supports arbitrary command execution, configurable timeout, and verbose logging, making it an operational RCE tool rather than a simple detector. The README mirrors the code behavior, explains usage, gives example commands such as whoami and cat /etc/passwd, and describes the vulnerability as a Python code-injection issue in the same API endpoint. No framework affiliation, persistence, lateral movement, or post-exploitation automation is present; the repository is focused narrowly on single-shot remote command execution against a specific vulnerable API surface.12-test-12 · Disclosed Mar 16, 2026 · python · markdown · networkOperationallangflow-exploitThis repository is a small standalone Python proof-of-concept exploit for CVE-2025-3248 affecting Langflow versions prior to 1.3.0. The repository contains only two files: a README describing the vulnerability and usage, and a single executable Python script, cve-2025-3248.py, which is the main exploit entry point. The exploit targets Langflow's unauthenticated code validation API endpoint at /api/v1/validate/code. It uses crafted JSON requests containing malicious Python code in the "code" field. The injected code abuses execution during function default-argument evaluation to run attacker-controlled Python statements on the server. The script supports two offensive capabilities: (1) command execution mode, which runs an arbitrary shell command via subprocess.check_output(..., shell=True) and attempts to recover command output from the application's returned error structure; and (2) reverse shell mode, which causes the target to open an outbound TCP connection to an attacker-controlled host and port, remaps stdin/stdout/stderr to that socket, and spawns /bin/sh using pty.spawn for interactivity. Operationally, the script disables TLS certificate verification, suppresses urllib3 SSL warnings, and uses a requests.Session with simple browser-like and JSON headers. It is not a scanner or detector; it is a direct exploitation tool. The code is functional and usable as-is, but payload customization is manual through command-line arguments rather than framework-driven, making the maturity best classified as OPERATIONAL rather than WEAPONIZED.preemware · Disclosed Aug 9, 2026 · python · markdown · web · networkOperationalCVE-2025-3248-Langflow-RCEThis is a small standalone Python proof-of-concept repository for CVE-2025-3248, an unauthenticated remote-code-execution issue affecting Langflow versions earlier than 1.3.0. It contains one executable script (exploit.py), a README describing the vulnerability and usage, and a .gitignore. The script uses requests to POST JSON to the target-supplied Langflow /api/v1/validate/code endpoint. Its code field embeds a Python decorator containing exec(...) and a compact reverse-shell payload. That payload attempts to connect back to an attacker-supplied TCP listener, attach file descriptors 0/1/2 to the socket, and spawn sh in a pseudo-terminal. Command-line arguments configure the target URL, callback host, callback port (default 9001), and HTTP timeout. It does not perform vulnerability verification, target discovery, persistence, credential theft, or post-exploitation actions; it simply reports the HTTP response and instructs the operator to check the listener.hideki233 · Disclosed Aug 28, 2026 · python · web · networkOperationalEXPLOIT-CVE-2026-9198This six-file repository is a standalone Python proof-of-concept and deliberately vulnerable Docker Compose lab for a claimed IBM Langflow OSS unauthenticated RCE chain. The main entry point, exploit.py, uses requests and exposes CLI options for a target URL, arbitrary command, optional supplied bearer token, TLS-verification bypass, request timeout, and configurable reverse-shell callback. It first requests /api/v1/auto_login to retrieve a purported SUPERUSER access token, then POSTs attacker-controlled Python source to /api/v1/validate/code using that token. The crafted source abuses evaluation of a function default argument to execute Python during validation. Command mode invokes a base64-wrapped shell command and intentionally raises an exception containing captured output, which the tool extracts from function.errors. Reverse-shell mode connects back to an operator-defined TCP listener and launches /bin/sh -i. docker-compose.yml creates a local-only lab with langflowai/langflow:1.10.0, PostgreSQL 16, LANGFLOW_AUTO_LOGIN=true, and Langflow published only at 127.0.0.1:9999. README.md provides exploitation, lab setup, mitigation, and version claims; requirements.txt contains only requests>=2.28. No external repository URL, git reference, or archive size was supplied in the analyzed content.joaovicdev · Disclosed Aug 29, 2026 · python · yaml · network · webOperationalLangflow-RCE-CVE-2025-3248This is a seven-file standalone Python exploit suite, not a known exploit-framework module. The executable entry point, exploit.py, parses a target URL and selects either command execution, reverse-shell delivery, or scanning. modules/exploit.py constructs a requests Session and POSTs JSON to the target-derived /api/v1/validate/code path; its code field evaluates __import__('os').system(...) to execute a supplied command. modules/revshell.py wraps that primitive in a Python socket one-liner that connects back to a configurable listener and invokes /bin/bash -i. modules/scanner.py sends an echo-based payload and treats HTTP 200 or reflected VULNERABLE text as an indication of exposure. utils.py provides logging/banner helpers, and requirements.txt lists requests and colorama. The command POST is made before response handling, so exploitation can still reach the target endpoint. However, successful command and scanner paths call logger.success(), which is not a method on Python's standard logging.Logger. This raises AttributeError: execute catches it and reports a connection-style error/returns no output, while scanner catches it and returns false. Consequently, successful exploitation or detection may occur but be incorrectly reported and command output will not be presented. The reverse-shell payload is nevertheless dispatched before this local reporting bug occurs.LeotheGGman · Disclosed Sep 5, 2026 · python · networkOperationalCVE-2025-3248The repository contains a standalone Python exploit PoC, documentation, an MIT license, a generic security-policy template, and a GitHub CodeQL workflow. `CVE-2025-3248.py` is the only exploit entry point. It accepts a target and command, suppresses HTTPS certificate validation, requests an unauthenticated JWT from Langflow's auto-login route, lists flows, locates a `code_runner`/PythonREPL flow, injects Python command-execution and reconnaissance logic, patches the flow, and triggers its execution. It is an operational exploit rather than a scanner because it changes target state and executes an operator-controlled command. The README characterizes CVE-2025-3248 as unauthenticated code execution via `/api/v1/validate/code` in Langflow versions before 1.3.0; however, the visible exploit implementation uses an auto-login plus flow-modification chain instead of directly calling that documented validation endpoint.zoly-zoly · Disclosed Sep 9, 2026 · python · yaml · web · networkOperationalCVE-2025-3248-CTF-LabThis is a standalone CTF challenge and working exploit reproducing the code-validation flaw associated with CVE-2025-3248. It does not install actual Langflow: a small FastAPI application advertises Langflow Core API version 1.2.0 and recreates the vulnerable validation behavior. The README describes Langflow versions before 1.3.0 as affected; the supplied implementation specifically emulates 1.2.0. The validator parses submitted Python into an AST and executes individual import and function-definition nodes using compile() and exec(). Executing a function definition evaluates decorators and default arguments even without calling its body. The exploit abuses a decorator to execute commands and leaks their output through returned exception messages. The 10-file repository contains exploit.py, a Python standard-library HTTP client; chall/app/main.py, the vulnerable API; an HTML/CSS/JavaScript editor that can also submit payloads; a shell entrypoint; a Dockerfile; pinned FastAPI, Uvicorn, and Pydantic dependencies; Docker Compose configuration; documentation; and editor/ignore configuration. Five files contain application, script, frontend, or Docker build code. The entrypoint writes an environment-supplied flag into a randomly named, world-readable root-directory file, creates a hint file, removes flag-related environment variables, and starts the service as appuser. The automated exploit makes two requests: one to enumerate root-directory filenames and another to read the matching flag file. Its command-execution helper is reusable, but the shipped command sequence is hardcoded and does not provide a reverse shell, persistence, privilege escalation, or external exfiltration destination. The Dockerfile's rm -rf command only cleans package-manager cache and is not evidence of a fake exploit. No repository URL, git reference, or archive size was supplied; the empty metadata fields and zero size are placeholders for unavailable information.cpwjdgh · Disclosed Oct 18, 2026 · python · shell · web · networkOperationalCVE-2025-3248-POCThis repository provides a proof-of-concept (POC) exploit for CVE-2025-3248, targeting the 'Langflow' application. The main exploit is implemented in 'poc.py', which allows an attacker to execute arbitrary system commands on a vulnerable Langflow instance by sending a crafted Python payload to the '/api/v1/validate/code' API endpoint. The payload writes a message file to the target's filesystem and executes a shell command provided by the attacker, returning the output. The repository also includes 'patch.py', a script intended to 'patch' the vulnerability by overwriting the vulnerable function on the target system, also via code injection. The structure is straightforward, with two main Python scripts (exploit and patch), a README describing usage, and standard project files. The exploit demonstrates operational maturity, as it provides a working payload and clear instructions for use.PuddinCat · Disclosed Apr 10, 2025 · python · networkPoCCVE-2025-3248This repository contains a proof-of-concept exploit for CVE-2025-3248, a critical unauthenticated remote code execution vulnerability in Langflow AI. The main exploit script (CVE-2025-3248.py) is a Python 3 tool that provides an interactive shell to the attacker, allowing arbitrary system commands to be executed on the target server. The exploit works by sending a specially crafted Python function to the /api/v1/validate/code endpoint, which is vulnerable to code injection. The script disables SSL verification for convenience and uses the requests and colorama libraries. The repository includes a README with usage instructions, a requirements.txt for dependencies, and a standard MIT license. No detection or fake code is present; this is a functional exploit PoC targeting a real vulnerability.0xgh057r3c0n · Disclosed Jun 18, 2025 · python · networkPoCMass-CVE-2025-3248This repository contains a Python exploit script (Mass-CVE-2025-3248.py) and a README.md. The exploit targets CVE-2025-3248, a critical unauthenticated remote code execution vulnerability in Langflow versions prior to 1.3.0. The vulnerability exists in the '/api/v1/validate/code' API endpoint, which improperly executes user-supplied code. The script allows the user to specify a single target URL or a file containing multiple targets, and sends a crafted POST request to the vulnerable endpoint with a Python payload that executes arbitrary shell commands on the server. The output of the command is returned to the attacker. The exploit is operational, providing a working RCE payload, and is intended for use in penetration testing or vulnerability validation. The README provides usage instructions and highlights the critical nature of the vulnerability.dennisec · Disclosed Jun 23, 2025 · python · markdown · networkPoC
Exposure surface

Affected products & vendors

Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.

VendorProductType
LangflowLangflowapplication
FAQ

Frequently asked questions

Is CVE-2025-3248 exploited in the wild?
Yes. As of Nov 17, 2026, Mallory has reports of exploitation in the wild, and CISA added it to KEV on May 5, 2025.
Is there a public exploit for CVE-2025-3248?
Yes. Mallory tracks 27 public exploits or proofs of concept for CVE-2025-3248, the earliest disclosed Apr 9, 2025.
Is there a patch for CVE-2025-3248?
Upgrade the langflow package to version 1.3.0 or later and langflow-base to version 0.3.0 or later, as applicable. Langflow 1.3.0 adds an active-user authentication dependency to the vulnerable endpoint.
What products are affected by CVE-2025-3248?
Langflow.
How severe is CVE-2025-3248?
CVE-2025-3248 has a CVSS v3.1 score of 9.8 (Critical) and an EPSS score of 100.0% (100th percentile) as of Nov 17, 2026.
Activity feed

Recent activity

289 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.

View more in app →
safebreach blogNews · Nov 16, 2026AI Security Validation: Prove Your Defenses Hold | SafeBreachA critical unauthenticated remote code execution vulnerability in Langflow versions before 1.3.0, rated CVSS 9.8. An exposed code validation endpoint allows attackers to execute code on the server. The article states that CISA includes it in the Known Exploited Vulnerabilities catalog and describes testing servers for exploitability. Recommended defenses include upgrading to version 1.3.0 or later, restricting network exposure, requiring authentication, validating network and web application controls, and monitoring for suspicious processes and outbound connections.Read more ↗security affairsNews · Nov 16, 2026ARTEX: How Cybercriminals Use AI Agents to Scale Their AttacksA known Langflow vulnerability sought by ARTEX-driven AI agents during a campaign targeting AI-service credentials. The article does not describe the flaw's technical mechanism or establish successful exploitation; Amazon assessed campaign records rather than observing the attacks directly.Read more ↗the hacker newsNews · Nov 16, 2026Amazon Uncovers AI Agent Campaign Targeting LiteLLM, Dify, and Other Apps for API KeysAn unauthenticated remote code execution vulnerability affecting Langflow versions before 1.3.0. It appears in the captured agents' attack playbook, although Amazon could not verify their reported outcomes. Separately, CISA added it to KEV on May 5, 2025, and identifies it as used in ransomware campaigns. Version 1.3.0 fixes the flaw.Read more ↗vulncheck blogNews · Nov 15, 2026Exploitation Flows Targeting Japan | Blog | VulnCheckA Langflow vulnerability associated with 210 exploitation hits from 30 unique IPs. Observed activity included reconnaissance and Sliver command-and-control.Read more ↗
+96 more in the timeline. Sign up to see them
Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Mallory platform

The version that knows your environment.

This page is what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which adversaries are exploiting it right now, and which detections to deploy tonight.
Exposure mappingQuery your assets running an affected version, and investigate the blast radius.
Threat actor evidence31Every report linking this CVE to a named adversary.
Associated malware67Malware families linked to this CVE, with evidence and IOCs.
Detection signatures3YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Vendor-by-vendor mappingCross-references every affected SKU, including bundled OEM variants.
Social activity96Community discussion across Reddit, Mastodon, and other social sources.