The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions to terminate system processes using the kill command based on port activity monitored by lsof.
[COMMAND_EXECUTION]: The skill facilitates privilege escalation within containers by providing commands to execute as the root user using docker compose exec -u root.
[COMMAND_EXECUTION]: The skill includes destructive commands for deleting Docker resources, including containers, images, and volumes, via docker system prune -a --volumes.
[COMMAND_EXECUTION]: The skill provides methods for obtaining interactive shell access (sh) to both running and failed containers.
[DATA_EXFILTRATION]: The skill reads container logs and configuration data through docker compose logs and docker compose config, which may expose sensitive information such as environment variables, secrets, or application data.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from container logs and inspection metadata without sanitization.
Ingestion points: docker compose logs, docker inspect.
Boundary markers: Absent.
Capability inventory: Process termination (kill), root execution (exec -u root), shell access (sh), and resource deletion (prune).
Sanitization: Absent.

troubleshoot