The Agent Skills Directory

COMMAND_EXECUTION (LOW): Scripts like analyze-build-failure.sh and inspect-container.sh execute local docker and aws CLI commands to perform diagnostics, which is the primary purpose of the skill.
DATA_EXFILTRATION (LOW): inspect-container.sh retrieves environment variables and logs from containers. This may expose secrets or configuration data to the agent if they are stored in the environment.
INDIRECT_PROMPT_INJECTION (LOW): The skill processes untrusted data from Docker build logs and container metadata. * Ingestion points: scripts/analyze-build-failure.sh (log file content), scripts/inspect-container.sh (container logs and metadata). * Boundary markers: None; data is output directly. * Capability inventory: Access to docker and aws commands. * Sanitization: None; output is raw text.
DYNAMIC_EXECUTION (LOW): scripts/check-platform.sh generates a temporary Dockerfile and executes docker build for diagnostic verification using a fixed template.

debugging-docker