The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructions direct the agent to execute a remote installation script by downloading it from https://edgespeak.com/install.sh and piping it directly to the shell (sh) if the CLI tool is not found. This pattern allows for unverified arbitrary code execution from a third-party source.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download of an installation script from an external domain (edgespeak.com) and notes that the system may perform automatic downloads of machine learning models from remote servers during initial execution.
[COMMAND_EXECUTION]: The skill instructs the agent to run various local shell commands using the edgespeak-cli tool for checking status, authenticating, and performing transcription tasks, granting the agent direct interaction with the host's command-line interface.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection and command injection because it interpolates user-supplied media file paths and transcription parameters directly into shell command strings without validation or sanitization. Evidence chain: 1. Ingestion points: media path input and subtitle parameters (file: SKILL.md); 2. Boundary markers: Absent; 3. Capability inventory: subprocess execution of edgespeak-cli (file: SKILL.md); 4. Sanitization: Absent.

edgespeak-transcribe