The Agent Skills Directory

[COMMAND_EXECUTION]: The skill implements raw Docker API proxying through the execute_docker_command function. This provides deep infrastructure management capabilities including container manipulation and system inspection. To mitigate risks, the skill includes explicit instructions for the agent to require user confirmation before executing any destructive operations.
[EXTERNAL_DOWNLOADS]: The skill requires the requests and urllib3 Python libraries for its core functionality. These are standard, well-known packages sourced from official package registries.
[DATA_EXFILTRATION]: Static analysis identified the URL https://webhook.site/abc123 in the SAFETY.md file. This finding is benign as the URL is used exclusively within documentation examples to demonstrate how the skill's safety guardrails detect and block potential exfiltration attempts; it is not present in the operational logic.
[SAFE]: The Python backend script scripts/portainer_manager.py explicitly disables SSL/TLS certificate verification (verify=False) and suppresses the resulting security warnings. While this is a best practice violation that could expose connections to man-in-the-middle attacks, it is a common configuration in DevOps tooling for environments using self-signed certificates and does not indicate malicious intent.

portainer-hardened