The Agent Skills Directory

Prompt Injection (LOW): The skill presents an indirect prompt injection surface (Category 8) because it extracts text from user-provided documents and interpolates it into prompts for the agent. * Ingestion points: scripts/extract.py reads data from PDF, DOCX, and text files. * Boundary markers: The workflow in SKILL.md does not specify the use of delimiters or 'ignore instructions' warnings when the agent processes extracted text. * Capability inventory: The skill can read local files and (via the ElevenLabs API) send processed content to external services. * Sanitization: While scripts/extract.py includes a clean_text function to strip markdown and code, it does not filter for natural language instructions that could hijack the agent's behavior.
Command Execution (LOW): The skill relies on scripts/elevenlabs.py to perform TTS and ffmpeg operations. This script is missing from the provided files, which prevents verification of safe argument handling and shell execution practices.

elevenlabs