snmp-pentesting
Warn
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download third-party tools and resources from external sources.
- Fetches the snmp-shell tool from https://github.com/mxrch/snmp-shell.
- Downloads MIB files using apt-get install snmp-mibs-downloader and download-mibs.
- [PRIVILEGE_ESCALATION]: Administrative privileges are requested to install software and dependencies.
- Uses sudo apt install for installing system packages.
- Uses sudo python3 -m pip install to install Python dependencies for the downloaded snmp-shell tool.
- [COMMAND_EXECUTION]: The skill relies heavily on executing various CLI networking and security tools.
- Executes nmap for port scanning and service detection.
- Uses snmpwalk, snmpbulkwalk, snmpset, snmp-check, and braa for SNMP interaction.
- Invokes hydra and onesixtyone for credential brute-forcing.
- [REMOTE_CODE_EXECUTION]: The skill facilitates the execution of downloaded third-party code.
- Directs the agent to run python3 shell.py from the snmp-shell repository.
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through tool outputs.
- Ingestion points: Processes output from nmap, snmpwalk, and other SNMP tools which reflect data from remote network targets (SKILL.md, references/snmp-techniques.md).
- Boundary markers: No specific delimiters or instructions are provided to the agent to treat tool output as untrusted or to ignore embedded instructions.
- Capability inventory: Possesses capabilities for further command execution (snmpset, hydra, python) based on information extracted from tool outputs.
- Sanitization: No sanitization or validation of the data received from remote targets is performed before it influences subsequent agent actions.
Audit Metadata