The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent on using the kamal CLI to manage remote applications, including the kamal app exec command which allows running arbitrary commands within app containers on servers. This is a core administrative feature of the documented tool.
[DATA_EXPOSURE]: To function correctly, the agent is directed to read configuration and secret files such as config/deploy.yml and .kamal/secrets to understand the deployment environment and host mapping.
[EXTERNAL_DOWNLOADS]: The skill documentation includes links to official resources at kamal-deploy.org for additional command details, configuration options, and documentation. These are well-known official sources for the tool.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes external data from application logs (kamal app logs) and server audit trails (kamal audit).
Ingestion points: Project configuration files (config/deploy.yml), secrets (.kamal/secrets), application logs, and server audit logs.
Boundary markers: None identified in the instructions.
Capability inventory: Execution of various kamal subcommands via the shell, including those that modify server state or execute remote code.
Sanitization: No explicit sanitization or validation of the ingested log or configuration data is mentioned.

app