The Agent Skills Directory

[SAFE]: The skill acts as a reference for shell configuration best practices. It contains no malicious logic and focuses on implementing standard developer environment setups using native Fish and Zsh syntax.\n- [COMMAND_EXECUTION]: The skill instructions utilize diagnostic tools like grep and rg to audit configuration files (e.g., ~/.config/fish/conf.d/*.fish) for syntax errors or deprecated patterns. This usage is restricted to local user configuration directories for maintenance purposes.\n- [DYNAMIC_EXECUTION]: Provides patterns for initializing development tools like Starship, FNM, and Pyenv using eval hooks (e.g., eval \"$(starship init zsh)\"). These are documented industry-standard integration methods and include safety guards using type -q or $+commands to verify tool presence.\n- [INDIRECT_PROMPT_INJECTION]: The skill identifies user configuration files (~/.zshrc, ~/.zshenv, ~/.config/fish/config.fish) as primary ingestion points. It lacks specific boundary markers for these files but provides a capability inventory limited to shell management tools (Read, Write, Grep, Bash) and mandates sanitization via syntax validation (fish -n, zsh -n) before applying changes.

shell-config