Exploit catalog results

Showing 3 PoCs on this page

GitHub

sonnelon/CVE-2026-33017-PoC

Repository PoCStars: 0Created 2026-08-13
ExploitCVE-2026-330171 file

3.7 KiB

Analysisdeepseek-v4-pro:cloud ·

Technical assessment

A Python script that exploits CVE-2026-33017 in Langflow by sending a crafted POST request to the unauthenticated payload withheld endpoint. The payload contains a custom component with embedded Python code that executes a reverse shell via os.system().

Backdoor review

No backdoor observed in reviewed code

The PoC is a straightforward Python script that exploits CVE-2026-33017 by sending a crafted POST request to a target Langflow instance. The payload contains a reverse shell connecting back to the operator-supplied host and port. All behavior is consistent with the disclosed unauthenticated RCE exploit, and no concealed or deceptive actions targeting the operator were found.

ClassificationExploit
Model confidence100%
AuthenticationNot required
Languagespython
Target softwarelangflow
Attack typesremote code executioncode injectioneval injection
Evidence & reasoningClassification basis · observed behavior · safety review
Technical evidence

Classification basis and observed behavior

Classification basis

The script actively sends a crafted payload to a vulnerable endpoint to achieve remote code execution via a reverse shell, which is the definition of an exploit.

exploit.py:1-110

Requirements

  • Target Langflow instance with a public flow IDexploit.py:9
  • Attacker-controlled listener to receive reverse shellexploit.py:10-11

Observed behavior

  • Constructs a reverse shell payload using os.system() with a bash TCP reverse shell commandexploit.py:16-28
  • Embeds the reverse shell code inside a custom Langflow component definition in a JSON payloadexploit.py:30-83
  • Sends an unauthenticated POST request to the vulnerable build_public_tmp endpoint with the malicious payloadexploit.py:85-103
Safety-review evidence

Behaviors behind the backdoor verdict

Observables

Reverse Shell
Payload withheldThe exploit payload executes a reverse shell on the target, connecting back to the operator-controlled address and port provided via command-line arguments.exploit.py:19
Target Endpoint
Payload withheldThe script sends the exploit payload to this unauthenticated endpoint on the target server, which is the vulnerable endpoint described in CVE-2026-33017.exploit.py:86
Review boundaries

What the analysis did not establish

  • The review is based solely on the provided text of exploit.py; no other files or runtime behavior were examined.
Model interpretation

This review is limited to the supplied PoC code and context. It does not assert that the code works or is safe to execute.

GitHub

sonnelon/CVE-2026-23744-PoC

Repository PoCStars: 0Created 2026-08-10
ExploitCVE-2026-237443 files

1.7 KiB

Analysisdeepseek-v4-pro:cloud ·

Technical assessment

Python script that sends a crafted HTTP POST request to a vulnerable MCPJam inspector endpoint to trigger installation of an MCP server, resulting in a reverse shell via netcat.

Backdoor review

No backdoor observed in reviewed code

The PoC exploit.py sends a crafted HTTP POST to the target's payload withheld endpoint with a serverConfig that instructs the target to execute 'payload withheld', establishing a reverse shell. This behavior directly matches the described CVE-2026-23744 remote code execution vulnerability. No concealed, unrelated, or operator-directed harmful behavior beyond the stated exploit was observed.

ClassificationExploit
Model confidence95%
AuthenticationNot required
LanguagesPython
Target softwareMCPJam inspector
Attack typesRemote Code ExecutionReverse Shell
Evidence & reasoningClassification basis · observed behavior · safety review
Technical evidence

Classification basis and observed behavior

Classification basis

The script actively sends a malicious payload to the target to achieve remote code execution via a reverse shell, which is the definition of an exploit.

exploit.py:26-45

Requirements

  • Target running vulnerable MCPJam inspector (<= 1.4.2) with the payload withheld endpoint exposed.exploit.py:27
  • Attacker must have a netcat listener running on a reachable IP and port to receive the reverse shell.exploit.py:8-9

Observed behavior

  • Sends an unauthenticated HTTP POST request to payload withheld with a JSON payload specifying a server command (busybox nc) that connects back to the attacker's IP and port, executing /bin/bash.exploit.py:27-45
  • Checks if the response status code is not 404 to determine success.exploit.py:46-47
Safety-review evidence

Behaviors behind the backdoor verdict

Observables

Reverse Shell
Payload withheldThe PoC payload instructs the vulnerable MCPJam inspector to spawn a reverse shell to the attacker-controlled IP and port, consistent with the CVE description of RCE via crafted HTTP request.exploit.py:31-37
Target Endpoint
Payload withheldThe HTTP endpoint targeted by the exploit, matching the vulnerability description of an unauthenticated MCP server installation endpoint.exploit.py:27
Review boundaries

What the analysis did not establish

  • One file (README.md) is omitted from the text evidence; its content is unknown.
  • The evidence does not include the execution output or verification that the exploit works.
  • One file (unclassified, 6 bytes) was omitted from text analysis; its metadata-only presence does not indicate backdoor behavior but prevents full artifact inspection.
  • Binary files were not analyzed; none were identified in the artifact.
Model interpretation

This review is limited to the supplied PoC code and context. It does not assert that the code works or is safe to execute.

GitHub

sonnelon/CVE-2025-59528-PoC

Repository PoCStars: 0Created 2026-08-10
ExploitCVE-2025-595283 files

5.1 KiB

Analysisdeepseek-v4-pro:cloud ·

Technical assessment

Python script that sends a crafted HTTP POST request to a Flowise instance to achieve remote code execution via a reverse shell. The payload exploits a code injection vulnerability in the mcpServerConfig parameter by using the Function() constructor to execute a shell command.

Backdoor review

No backdoor observed in reviewed code

The PoC demonstrates the documented CVE-2025-59528 remote code execution vulnerability against Flowise 3.0.5. The exploit payload sends a reverse shell command to the target server, which is the expected behavior for a PoC of this CVE. No concealed backdoor, unrelated payload, or operator-directed harm beyond the stated exploit was observed.

ClassificationExploit
Model confidence98%
AuthenticationRequired
Languagespython
Target softwareFlowise
Attack typesremote code executioncode injection
Evidence & reasoningClassification basis · observed behavior · safety review
Technical evidence

Classification basis and observed behavior

Classification basis

The artifact contains a Python script (exploit.py) that actively sends a malicious payload to a vulnerable endpoint to execute arbitrary commands on the target server, which is the definition of an exploit.

exploit.py:16exploit.py:30-31

Requirements

  • Requires a valid API key for the target Flowise instance.exploit.py:35
  • Requires a network listener (e.g., netcat) to receive the reverse shell.README.md:44-47

Observed behavior

  • Constructs a JSON payload containing a reverse shell command inside the mcpServerConfig field.exploit.py:12-19
  • Sends an HTTP POST request to the /api/v1/node-load-method/customMCP endpoint with the payload and an Authorization header.exploit.py:24-31
  • Checks the HTTP response status code to report success or failure.exploit.py:44-47
Safety-review evidence

Behaviors behind the backdoor verdict

Observables

Reverse Shell Payload
Payload withheldThe payload creates a reverse shell connection back to the attacker-controlled IP and port, which is the intended demonstration of remote code execution for CVE-2025-59528.exploit.py:16
Review boundaries

What the analysis did not establish

  • One file (likely a screenshot or image) was omitted from the text evidence and not analyzed.
  • The evidence does not include the execution output or verification that the exploit works.
  • One file (metadata only) was omitted from the evidence and not reviewed.
  • Binary content was not analyzed; the artifact contained no identified binary files.
Model interpretation

This review is limited to the supplied PoC code and context. It does not assert that the code works or is safe to execute.