steganography

Fail

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: CRITICALDATA_EXFILTRATIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONPRIVILEGE_ESCALATION
Full Analysis
  • [DATA_EXFILTRATION]: The skill includes practical implementations for establishing covert communication and exfiltration channels.
  • network-protocol-steganography-guide.md contains Python scripts using the scapy library to embed data in ICMP echo request payloads and IP header TOS/DSCP fields.
  • The guide also provides code for DNS tunnel steganography and TCP timing channels, which are specialized techniques used to bypass network monitoring and exfiltrate data.
  • [EXTERNAL_DOWNLOADS]: The instructions direct the user to install and use a large number of third-party tools and libraries from various external sources.
  • References tools such as steghide, zsteg, binwalk, stegcracker, stegseek, jsteg, and outguess hosted on platforms like SourceForge and GitHub.
  • Mentions the external security resource book.hacktricks.xyz, which, while a common security reference, is often flagged by automated reputation scanners.
  • [DYNAMIC_EXECUTION]: Several guides utilize dynamic code execution to perform complex operations.
  • audio-video-steganography-guide.md and network-protocol-steganography-guide.md contain multi-line Python snippets executed via python3 -c for media encoding and network protocol manipulation.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, potentially malicious files which could influence agent behavior.
  • Ingestion points: The agent is instructed to read and analyze untrusted images, audio files, and PCAP captures using tools like exiftool, binwalk, and zsteg.
  • Boundary markers: There are no explicit delimiters or instructions to the agent to ignore potentially malicious content embedded within the files being analyzed.
  • Capability inventory: The skill provides access to powerful tools like Bash, Write, and WebFetch, which could be abused if the agent follows instructions hidden within a steganographic carrier.
  • Sanitization: There is no mechanism described for sanitizing the output of extraction tools before the agent processes the resulting data.
  • [COMMAND_EXECUTION]: The skill relies on the execution of powerful CLI utilities for file system interaction and network analysis.
  • Examples include binwalk, foremost, and steghide, which perform deep analysis of binary structures and metadata.
  • [PRIVILEGE_ESCALATION]: The payloads.md guide suggests installing software using apt, which typically necessitates administrative privileges.
Recommendations
  • Contains 2 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Sep 19, 2026, 04:36 PM