Read the researcher's report on HackerOne
The report identifies an unsafe YAML deserialization call in a source-code file and states that loading attacker-controlled YAML could enable code execution. It does not show that untrusted input reaches the call or provide a working demonstration.
How it was found
- The reporter identified a YAML-loading call in a testing-related module.
- They noted that the loader can construct arbitrary Python objects.
- They proposed safe loading, but did not test an attacker-controlled input or confirm execution.
- Access needed
- Not clear from the report
- User interaction
- Not clear
- Whose object
- No object involved
- Boundary crossed
- Not established; the report does not show an attacker-controlled path to the loader.
- Root cause
- A general-purpose YAML loader is used where untrusted YAML could be processed.
- Impact
- If an attacker can supply input to this call, deserialization may execute code, but the report does not establish that exposure.
- Fix
- Replace the general-purpose loader with a safe YAML loader.
- Components
- YAML parser, testing module
- Confidence
- low
Unsafe deserialisationCode injection / eval
This analysis is CyberSIXT's own summary of the public report, written by a model and checked automatically. It may be wrong: the researcher's report is the source. Written 2026-11-15. Ask for a correction or removal.
| Report | Access | Whose object | Boundary crossed |
|---|---|---|---|
| Authenticated Code Execution through Phar deserialization in CSV Importer as Shop manager in WooCommerce | A user account | The attacker's own | A shop manager's import request caused server-side processing of an uploaded Phar file. |
| Remote code execution on rubygems.org | A user account | No object involved | A gem upload crossed from untrusted input into server-side code execution. |
| connect.8x8.com: Deserialization Vulnerability in Automation Builder via Jint→Newtonsoft serializer coercion (TypeNameHandling) | A user account | The attacker's own | An authenticated user crossed from configuring an HTTP request step into server-side .NET type deserialization. |
| Loading YAML in Java client can lead to command execution | A user account | No object involved | YAML input crossed into Java object deserialization and code execution. |
| RCE via exposed JMX server on jabber.37signals.com/jabber.basecamp.com | No account | No object involved | Unauthenticated access to an exposed JMX management service enabled potential remote code execution. |
| Remote Code Execution via Insecure Deserialization in Telerik UI (CVE-2019-18935) | No account | No object involved | Unauthenticated access to the application's Telerik upload handler allowed a test file to be uploaded. |
| Remote Code Execution (RCE) in a Sony Pictures WebSystem | Not clear from the report | No object involved | Not specified; alleged code execution on the web system. |
| Unsafe deserialization in Libera Pay allows to escalate a SQL injection to Remote Command Execution | A user account | Another user's | An invited user's notification rendering could deserialize context altered through a separate SQL injection. |
| Remote Code Execution via CVE-2019-18935 | No account | No object involved | An unauthenticated request to the exposed upload handler reached server-side deserialization and caused a delayed response. |
| Pickle deserialization vulnerability in XComs | A user account | No object involved | A DAG author caused poisoned XCom data to be deserialized by another task or an eligible web UI path. |
| Attacker can add arbitrary data to the blockchain without paying gas | A user account | The attacker's own | A peer's transaction parser accepted trailing data and propagated it into a mined block. |
| SQL Injection + Insecure Deserialization leads to Remote Code Execution on https://krisp.ai | No account | No object involved | Unauthenticated requests reached server-side SQL and deserialization processing, enabling code execution. |