The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's documentation includes a common installation pattern that pipes a remote script to bash (curl -fsSL https://servel.dev/install.sh | bash). While this is a sensitive operation, it targets the official domain of the service being managed by the skill and is documented as the standard installation method.
[COMMAND_EXECUTION]: The skill makes extensive use of CLI commands to manage servers, deployments, and infrastructure. This includes powerful operations such as servel ssh, servel exec, and modifying firewall rules via servel ban. These operations are well-documented and core to the skill's functionality.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests data from local project files that could contain untrusted content.
Ingestion points: Reads configuration and environment data from servel.yaml, .servel/state.json, and .env files within the project directory.
Boundary markers: None identified in the prompt templates to delimit data from instructions.
Capability inventory: The skill has high capabilities, including executing remote commands, managing secrets, and modifying infrastructure via the servel CLI.
Sanitization: No explicit sanitization or validation of the input file content is performed before interpolation into commands.

servel