Generate Protoss-style (StarCraft) voice effects using SoX and FFmpeg.
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: protoss-voice Version: 1.1.1 The skill bundle is designed to apply audio effects using `ffmpeg` and `sox`. The `protoss_fx.py` script correctly uses `subprocess.run` with lists of arguments, preventing shell injection vulnerabilities. Temporary files are handled and cleaned up appropriately. While the `SKILL.md` contains instructions for the agent to 'not narrate this step unless debugging' and 'clean up ... silently', these are related to controlling the agent's verbosity for user experience during a legitimate processing task, and do not instruct the agent to perform any malicious actions such as data exfiltration, unauthorized access, or persistence.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Using the skill allows the agent to run local media-processing binaries and create or replace generated audio outputs.
The skill invokes local ffmpeg and sox commands on the supplied audio file. This is central to the stated audio-effect purpose and uses argument lists rather than shell execution.
subprocess.run(["ffmpeg", "-y", "-i", input_path, "-vn", temp_wav_in], check=True, ...); subprocess.run(cmd_psi, check=True)
Use it only on intended audio files, keep backups if an existing _psionic output matters, and install ffmpeg/sox from trusted package sources.
The agent may perform expected processing and temporary-file cleanup without describing every step during normal use.
The workflow tells the agent to run the processor and silently clean temporary artifacts. This is scoped to the audio task, but it is still a local file-operation behavior users should notice.
Execute `protoss_fx.py` on the raw file. *Do not narrate this step unless debugging.* ... Clean up raw/intermediate files silently if they are temporary.
If preserving raw or intermediate audio is important, ask the agent not to delete temporary files before invoking the skill.