hunt-deserialization
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to clone a third-party exploitation repository which is not from an established or verified organization. Evidence:
git clone https://github.com/pimps/JNDI-Exploit-Kit. - [COMMAND_EXECUTION]: Provides numerous pre-formatted shell commands for probing targets and delivering payloads via network requests. Evidence: Extensive use of
curland CLI tools for target interaction. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from source code and server responses that could potentially contain malicious payloads targeting the analyzer. Ingestion points: Analyzing files via
grepand processing target header/cookie data. Boundary markers: A mandatory confirmation gate is included for user oversight. Capability inventory: Network access and execution of various language runtimes. Sanitization: No evidence of automated sanitization for ingested content. - [DYNAMIC_EXECUTION]: Utilizes runtime script execution and specialized security tools to dynamically generate serialized payloads. Evidence: Construction of Python pickle and Java serialized objects using inline code and external JAR files.
Audit Metadata