Chen Openai Whisper

Local speech-to-text with the Whisper CLI (no API key).

Security Scan

The skill is an instruction-only wrapper for a local Whisper CLI; its requirements (the whisper binary) and instructions are coherent and proportionate to the stated purpose.

Purpose & Capability

Name/description say 'local speech-to-text with Whisper CLI' and the skill only requires the 'whisper' binary and shows CLI usage. Requiring the whisper binary and offering a brew install is consistent with the claimed purpose.

Instruction Scope

SKILL.md instructs the agent to run the whisper CLI against local audio files and notes that models are downloaded to ~/.cache/whisper. The instructions do not ask for unrelated files, environment variables, or external endpoints. Note: the doc claims '--model' defaults to 'turbo', which is an unusual model name for classic Whisper distributions — you may want to verify supported model names for the specific brew package.

Install Mechanism

Install uses a Homebrew formula ('openai-whisper'), which is low-risk compared with arbitrary downloads. However, Homebrew formulas can come from different taps — verify the formula source (official Homebrew/core vs a third-party tap) before installing.

Credentials

No environment variables, credentials, or config paths are requested. This is proportionate for a local CLI transcription skill.

Persistence & Privilege

Skill is not always-enabled and does not request elevated or persistent privileges. It does cause models to be cached to ~/.cache/whisper on first run (disk usage), which is normal behavior for local model downloads.

Assessment

This skill is an instruction wrapper around a local Whisper CLI and looks coherent. Before installing: 1) Confirm which Homebrew tap provides the 'openai-whisper' formula (prefer official Homebrew/core or a trusted tap). 2) Expect large model files to be downloaded to ~/.cache/whisper on first use — ensure you have disk space and are comfortable with local storage of audio-derived models. 3) Verify supported model names for the package (the README's default 'turbo' is unusual for classic Whisper). 4) The skill will read any audio file you point it at and write transcripts locally — don't feed sensitive audio unless you trust the installed binary. Overall this appears consistent with its stated purpose.

Whisper (CLI)

Use whisper to transcribe audio locally.

Quick start

• whisper /path/audio.mp3 --model medium --output_format txt --output_dir .
• whisper /path/audio.m4a --task translate --output_format srt

Notes

• Models download to ~/.cache/whisper on first run.
• --model defaults to turbo on this install.
• Use smaller models for speed, larger for accuracy.