Exploit catalog results

Showing 2 PoCs on this page

GitHub

cyeezy08/Kimai-CVE-2026-49865-POC

Repository PoCStars: 0Created 2026-07-30
ExploitCVE-2026-49865CVE-2026-5282410 files

71.4 KiB

Analysisdeepseek-v4-pro:cloud ·

Technical assessment

The artifact is a complete exploit toolkit for CVE-2026-52824, a critical authentication bypass in Kimai. It includes a Python script (exploit.py) that forges a KIMAI_REMEMBER cookie using a hardcoded default APP_SECRET and an empty signature_properties hash, then uses the cookie to authenticate against a target instance and access protected pages.

Backdoor review

No backdoor observed in reviewed code

The reviewed text files (README.md, exploit.py, requirements.txt) describe and implement a proof-of-concept exploit for CVE-2026-52824, a known vulnerability in Kimai. The code forges a remember-me cookie using a hardcoded default secret and attempts to access protected pages. No concealed backdoor, credential exfiltration, persistence mechanism, or unrelated payload was observed. The behavior is consistent with the stated exploit purpose.

ClassificationExploit
Model confidence100%
AuthenticationNot required
LanguagesPythonMarkdown
Target softwareKimai
Attack typesauthentication bypassaccount takeovercookie forgery
Evidence & reasoningClassification basis · observed behavior · safety review
Technical evidence

Classification basis and observed behavior

Classification basis

The primary artifact is classified as an exploit because the main Python script (exploit.py) contains code that forges a valid authentication cookie and uses it to gain unauthorized access to a target application, which is the definition of an exploit. The README explicitly describes it as a 'Proof-of-concept exploit toolkit' and details the attack chain from advisory to admin session.

README.md:5README.md:15exploit.py:223-365

Requirements

  • Target must be a Kimai instance ≤ 2.57.0 using the default APP_SECRET (change_this_to_something_unique) as shipped in Docker or .env.dist.README.md:5exploit.py:8-11
  • The attacker must know or guess a valid username (e.g., admin, super_admin) and the corresponding user ID is sequential (admin = id:1).README.md:13README.md:42

Observed behavior

  • The exploit script computes a constant FieldsHash (SHA256 of empty string) and uses the known default APP_SECRET to compute an HMAC-SHA256 signature, then assembles a forged KIMAI_REMEMBER cookie.exploit.py:70-100exploit.py:103-130
  • The script sets the forged cookie in an HTTP session and attempts to access protected paths (e.g., /en/dashboard, /en/admin/user/) to verify successful authentication bypass.exploit.py:311-365
  • The README describes the toolkit as containing a single-target exploit, a multi-threaded batch exploitation script, a scanner, version probes, and a Shodan discovery script.README.md:56-65
Safety-review evidence

Behaviors behind the backdoor verdict

Observables

Exploit Behavior
Payload withheldThis is the normal, documented behavior of the PoC exploit.exploit.py:103-130exploit.py:223-365
Network Activity
Payload withheldThis is expected for a PoC that validates the vulnerability on a target instance.exploit.py:137-190exploit.py:327-360
File Write
Payload withheldThis is a convenience feature for saving the cookie, not a persistence mechanism.exploit.py:292-295
Review boundaries

What the analysis did not establish

  • Only 3 of 10 files in the repository are included as readable text; the other 7 files (including batch_pwn.py, quick_scan.py, deep_probe.py, probe_batch.py, shodan-scanner.py) are omitted, so the full toolkit behavior cannot be confirmed.
  • The evidence packet reports complete_artifact_coverage as false, indicating that not all files were analyzed.
  • The CVE records for CVE-2026-49865 and CVE-2026-52824 are absent from the acquired CVEList, so the official vulnerability description cannot be cross-referenced.
  • 7 files (53,755 bytes) were omitted from the evidence packet and not reviewed. Their content is unknown.
  • The analysis is limited to static review of the provided text; no dynamic execution was performed.
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

cyeezy08/DoS-Braces-3.03

Repository PoCStars: 0Created 2026-07-17
ExploitCVE-2024-40688 files

21.5 KiB

Analysisdeepseek-v4-pro:cloud ·

Technical assessment

The artifact is a Python script (poc_braces_dos.py) that programmatically triggers a denial-of-service vulnerability in the braces npm package (version 3.0.3) by generating a combinatorial explosion via comma-separated brace expansion. It executes Node.js code that calls braces.expand() with a crafted input, measures memory and time consumption, and demonstrates the crash/OOM condition.

Backdoor review

No backdoor observed in reviewed code

The supplied evidence consists of a README.md and a Python PoC script (poc_braces_dos.py) that demonstrate a denial-of-service vulnerability in the braces npm package. The PoC script installs braces@3.0.3 and runs Node.js code to trigger excessive memory consumption via comma-separated brace expansion. No backdoor, trojan, or deceptive payload targeting the person running the PoC was observed. The script's behavior is limited to demonstrating the claimed DoS vulnerability.

ClassificationExploit
Model confidence95%
AuthenticationNot required
LanguagesPythonJavaScript
Target softwarebraces (npm package)
Attack typesDenial of Service (DoS)Resource Exhaustion
Evidence & reasoningClassification basis · observed behavior · safety review
Technical evidence

Classification basis and observed behavior

Classification basis

The primary artifact is poc_braces_dos.py, which is a script that actively exercises the vulnerability by sending a malicious input to braces.expand() and observing the resulting resource exhaustion. This constitutes exploit code, as it is intended to trigger the vulnerability, not merely detect it.

poc_braces_dos.py:1-7poc_braces_dos.py:64-84

Requirements

  • Node.js and npm must be installed to run the script and install the vulnerable package.poc_braces_dos.py:1
  • The vulnerable braces package version 3.0.3 must be installed (the script attempts to auto-install it).poc_braces_dos.py:9-10

Observed behavior

  • The script auto-installs braces@3.0.3 if not already present.poc_braces_dos.py:20-45
  • It constructs a malicious input string by repeating the pattern '{a,b}' multiple times (e.g., 22 times for 110 characters).poc_braces_dos.py:66
  • It executes Node.js code that calls braces.expand() with the crafted input, causing the library to materialize millions of items in memory (e.g., 4,194,304 items for n=22).poc_braces_dos.py:64-84
  • It measures and reports the resulting memory consumption (e.g., +1,196,410KB) and execution time (e.g., 9723ms), demonstrating resource exhaustion.poc_braces_dos.py:88-93
  • For larger inputs (n=25), the script reports a TIMEOUT/OOM condition, indicating a successful denial of service.poc_braces_dos.py:96-98
Safety-review evidence

Behaviors behind the backdoor verdict

Observables

Command Execution
Payload withheldThe PoC script installs a specific version of the braces package to demonstrate the vulnerability. This is expected for a PoC and does not install unrelated or malicious packages.poc_braces_dos.py:38
Code Execution
Payload withheldThe PoC script executes Node.js code that calls braces.expand() with crafted input to trigger high memory usage. The executed code is fully visible and limited to the vulnerability demonstration.poc_braces_dos.py:50-52
Review boundaries

What the analysis did not establish

  • The evidence packet includes only two text files (README.md and poc_braces_dos.py) out of eight total files in the repository. The other six files (including findings.md, verdict.md, disclosure-report.md, patch.diff, email-draft.txt) are omitted, so their content is unknown.
  • The analysis is based solely on the provided text; the code was not executed, and the reported behavior is derived from the script's logic and embedded output examples.
  • Six files in the repository were not provided as text content and were not analyzed. Their names and contents are unknown.
  • Binary content, if any, was not inspected per the evidence envelope's binary policy.
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.