The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a Jenkins pipeline and setup scripts that execute various shell commands, including docker compose, curl for downloading agent binaries, and shred for secure file deletion.
[PRIVILEGE_ESCALATION]: The deployment workflow explicitly requires the Jenkins agent to run with sudo NOPASSWD permissions to perform system-level tasks such as writing secrets to tmpfs directories and managing Docker services.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface where untrusted data from an external SCM repository (like Python scripts or Dockerfiles) is executed in a high-privilege context without sanitization.
Ingestion points: Files checked out from the application repository (SKILL.md).
Boundary markers: No specific delimiters or safety instructions are used to prevent malicious code in the repository from influencing the agent's behavior during deployment.
Capability inventory: The pipeline has access to sudo, docker, and arbitrary shell execution (sh steps).
Sanitization: The skill does not implement validation or escaping for the repository content before execution.
[PERSISTENCE]: The instructions suggest creating a systemd service for the Jenkins inbound agent to ensure it remains active across reboots and automatically restarts on failure.

jenkins-agent-l3-deploy