The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides a command to install the Kamal tool by piping a remote script from kamal-deploy.org directly into the shell. This pattern is a known security risk as it executes unverified code from the internet.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-controlled configuration files (deploy.yml) and Dockerfiles. As it possesses extensive permissions (Bash, Write, Edit), malicious instructions embedded in these project files could influence the agent to perform unauthorized actions.
[COMMAND_EXECUTION]: The skill relies heavily on the Bash tool to manage servers, execute container commands, and run deployment hooks.
[EXTERNAL_DOWNLOADS]: Contains examples of post-deploy hooks that make external network requests (e.g., to notify.example.com), demonstrating that the skill's workflow expects and utilizes outbound network access.

kamal-coder