The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the shell execution prefix (!) throughout the SKILL.md file to automatically run commands and gather system state.
Evidence includes checks for tool availability (command -v mise, fnox --version) and project configuration validation.
[DATA_EXFILTRATION]: The skill accesses and displays potentially sensitive information from local configuration files and environment definitions.
Sensitive paths accessed: ~/.config/mise/config.toml and ~/.config/fnox/config.toml (both use head -10 to display content).
Environment exposure: The skill executes grep -A5 '^\[env\]' mise.toml, which can reveal environment variables and secrets not protected by redaction.
Identity exposure: Runs infisical user get to display current user session information.
[REMOTE_CODE_EXECUTION]: The skill contains patterns for downloading and executing code from remote URLs without verification.
Piped execution: Recommends curl https://mise.run | sh if the mise tool is missing.
Remote script execution: The file references/infisical-patterns.md includes curl -1sLf 'https://dl.infisical.com/get-cli.sh' | bash. Note: Infisical is a well-known service domain.
[EXTERNAL_DOWNLOADS]: Fetches installation scripts and configuration references from external domains including mise.run and dl.infisical.com.

devtools-secrets