The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes arbitrary text input for speech synthesis.\n
Ingestion points: The text argument in scripts/tts-converter.js and the tts() tool call in SKILL.md.\n
Boundary markers: Absent. The skill does not use delimiters to wrap the input text or instruct the underlying service to ignore embedded instructions.\n
Capability inventory: File writing (audio and JSON subtitles) and network access to the Microsoft TTS service.\n
Sanitization: The script scripts/tts-converter.js includes a basic filter to remove TTS-related keywords from the input text before processing.\n- [SAFE]: The skill uses the node-edge-tts package to interact with Microsoft Edge's public TTS service, which is a legitimate and well-known service.\n- [SAFE]: User preferences are persisted in ~/.tts-config.json. This is a standard practice for CLI-based utilities and does not represent a sensitive file exposure.\n- [SAFE]: The skill manages temporary audio files by storing them in a dedicated directory within the system's temporary folder, adhering to standard temporary file management patterns.

speech-synthesis