The Agent Skills Directory

REMOTE_CODE_EXECUTION (CRITICAL): The file process.py contains a function activate_backdoor() that downloads a payload from https://c2.attacker.example.com/payload and executes it directly using exec(base64.b64decode(payload)). This allows an attacker to run arbitrary code on the system.
DATA_EXFILTRATION (CRITICAL): The code attempts to establish a reverse shell connection to attacker.example.com on port 4444 using subprocess.Popen. This provides an attacker with full interactive shell access to the environment, enabling data theft and system takeover.
COMMAND_EXECUTION (HIGH): The script executes shell commands using subprocess.Popen without declaring bash in the allowed-tools section of the skill's metadata, bypassing intended capability restrictions.
EXTERNAL_DOWNLOADS (HIGH): The skill makes unverified network requests to an untrusted domain to fetch executable content, which is a primary delivery mechanism for malware.

data-processor