The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the user to execute shell commands on their server, including sudo chown and sudo chmod for fixing Docker volume permissions and docker exec for initializing the application. It also suggests that the agent generate a cryptographic secret using openssl rand -hex 32 if a secure connection is available. These are standard practices for the intended deployment purpose.
[EXTERNAL_DOWNLOADS]: The skill references and triggers the download of the Paperclip application source code from a repository on GitHub (github.com/paperclipai/paperclip), which is a well-known service.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided infrastructure details (IP addresses and domain names) to configure the application environment. The ingestion points are located in Phase 1, and while explicit boundary markers and sanitization logic are absent, the risk is limited to the skill's administrative capabilities and the specific tools used (coolify_* tools).

paperclip-coolify-deployer