The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local project files to determine the environment configuration. While this is standard for development tools, it represents a potential surface where malformed configuration files could influence agent behavior.
Ingestion points: The detect.sh script reads .ruby-version, .tool-versions, .mise.toml, and Gemfile from the project directory.
Boundary markers: No specific delimiters or safety instructions are used when passing version strings from these files to the agent.
Capability inventory: The agent is instructed to execute shell commands provided in the ACTIVATION_COMMAND output, followed by user-requested Ruby commands.
Sanitization: Basic format normalization is performed using tr and sed to extract version strings.
[DYNAMIC_EXECUTION]: The skill generates shell activation commands at runtime (such as eval "$(rbenv init -)" or sourcing manager-specific shell scripts) to establish the project's environment for the current session.
[COMMAND_EXECUTION]: The skill executes local bash scripts (detect.sh, detect-all-managers.sh, and set-preference.sh) and uses standard version manager binaries to probe the system state. It utilizes bash -lc to ensure login shell environment variables are respected during detection.

ruby-version-manager