The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local binaries (claude-speak-client and claude-speak-daemon) located within the user's home directory (~/Projects/claude-speak/). These commands are used to interact with a pre-installed text-to-speech system.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (shell injection) because user-supplied content is interpolated directly into a shell command as an argument.
Ingestion points: User-provided text intended for vocalization in SKILL.md commands.
Boundary markers: None specified to distinguish between text data and shell control characters.
Capability inventory: Execution of subprocesses via the command line.
Sanitization: No escaping or validation of the input string is mentioned before it is passed to the shell.

claude-speak