The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the user's local file system which could potentially contain malicious instructions.
Ingestion points: Reads configuration and documentation files from ~/dotfiles/.config/quickshell/, including AGENTS.md, Colors.qml, and colors.css.
Boundary markers: No specific boundary markers or instructions to disregard embedded commands in these files are present.
Capability inventory: The skill can execute shell commands (quickshell, hyprctl, git, pkill, setsid), run Python code, and open local files in the browser via xdg-open.
Sanitization: There is no explicit sanitization process for the data read from the local configuration files.
[COMMAND_EXECUTION]: The skill uses multiple shell commands to manage the Quickshell environment and verify the integrity of the user's desktop configuration.
Evidence: Uses pkill -x quickshell, setsid quickshell, hyprctl reload, and git apply --cached to manage processes, reload compositor rules, and stage code changes.
Purpose: These commands are standard for the development and lifecycle management of the targeted desktop environment components.
[DYNAMIC_EXECUTION]: The skill generates and executes code dynamically at runtime for the purpose of UI design and verification.
UI Prototyping: Creates temporary HTML/CSS/JS files in /tmp/ui-options/ and automatically opens them in the user's browser using xdg-open for manual verification.
Runtime Injection: Injects a custom JavaScript dumpUi function into the running Quickshell environment through an IPC call to inspect the live item tree.
Scripting: Executes a Python one-liner utilizing the fontTools library to verify the presence and names of font glyphs in system font files.

quickshell-modules