skills/mukul975/anthropic-cybersecurity-skills/performing-red-team-with-covenant/Gen Agent Trust Hub
performing-red-team-with-covenant
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/agent.pyis designed to facilitate remote command execution on 'grunts' (C2 agents) through the Covenant REST API. This behavior is the intended primary functionality for adversary simulation. - [CREDENTIALS_UNSAFE]: The Python agent in
scripts/agent.pyaccepts passwords via command-line arguments. This is considered unsafe practice as process arguments can be visible to other users on the system via process monitoring tools. - [PROMPT_INJECTION]: The skill processes data from a remote C2 server, creating an attack surface for indirect prompt injection.
- Ingestion points: The
list_listeners,list_grunts, andget_task_outputmethods inscripts/agent.pyretrieve data from the remote API. - Boundary markers: No delimiters or instructions are used to separate untrusted data from the agent's logic.
- Capability inventory: The agent performs network operations and tasks remote agents via the C2 API.
- Sanitization: The script does not validate or sanitize the JSON data received from the remote server before it is processed or printed.
Audit Metadata