The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the kitten @ send-text command to send keystrokes and commands to Kitty terminal windows. This allows the agent to execute arbitrary code or shell commands in any managed window or tab.
[DATA_EXFILTRATION]: The skill utilizes kitten @ get-text to retrieve the content of terminal windows, including the scrollback buffer. This provides a mechanism for the agent to access potentially sensitive information displayed in other terminal windows, such as environment variables, logs, or file contents.
[INDIRECT_PROMPT_INJECTION]: By reading terminal output via get-text, the skill introduces a surface for indirect prompt injection. Malicious instructions contained in the output of other processes, log files, or terminal-based applications could influence the agent's behavior.
Ingestion points: Terminal output buffers read via kitten @ get-text (SKILL.md).
Boundary markers: None identified in the provided instructions to distinguish between terminal data and agent instructions.
Capability inventory: Uses Bash to execute kitten @ commands for terminal manipulation (SKILL.md).
Sanitization: No sanitization or filtering of the retrieved terminal text is described.

kitty