← Back to Skills Marketplace
40
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install openai-whisper-hardened
Description
Local speech-to-text with the Whisper CLI (no API key).
Usage Guidance
This skill is internally consistent with local Whisper CLI usage. Before installing, verify the Homebrew formula (openai-whisper) comes from a trusted tap and inspect its contents if possible. Expect large model downloads to ~/.cache/whisper and ensure you have enough disk space and network bandwidth. The SKILL.md includes strict guardrails (no automatic network upload of transcripts, confirm consent for sensitive system paths) — those are good; follow them. If you need centralized/cloud storage of transcripts, handle uploads manually after review and consent rather than relying on the skill to perform network transmission.
Capability Analysis
Type: OpenClaw Skill
Name: openai-whisper-hardened
Version: 1.0.0
The skill bundle provides a security-hardened implementation of the OpenAI Whisper CLI for local speech-to-text. Both SKILL.md and SAFETY.md contain explicit, defensive instructions (guardrails) designed to prevent the AI agent from exfiltrating data, accessing sensitive system paths like /etc/, or deviating from the CLI scope. No malicious logic, data exfiltration, or harmful prompt-injection attempts were found; the content is focused on enhancing the agent's safety profile.
Capability Tags
Capability Assessment
Purpose & Capability
Name/description, required binary ('whisper'), and brew install (openai-whisper -> provides 'whisper') are coherent for a local Whisper CLI transcription skill.
Instruction Scope
SKILL.md limits actions to running the whisper CLI, documents model cache location, and includes explicit guardrails (no network exfiltration, confirm before transcribing sensitive system paths, prefer CLI over Python API). It does not instruct reading unrelated files or sending data to external endpoints.
Install Mechanism
Install uses a Homebrew formula (openai-whisper) which is a reasonable, low-risk install mechanism compared to arbitrary downloads. User should still validate the formula/tap before installing, but the mechanism itself is proportionate to the skill.
Credentials
The skill declares no environment variables, no credentials, and no config paths — consistent with a purely local CLI transcription tool.
Persistence & Privilege
Skill is not always-enabled, does not request elevated or cross-skill configuration, and does not request persistent credentials or system-wide changes.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install openai-whisper-hardened - After installation, invoke the skill by name or use
/openai-whisper-hardened - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of openai-whisper-hardened.
- Provides local speech-to-text transcription using the Whisper CLI (no API key needed).
- Includes quick start usage examples for transcription and translation tasks.
- Implements security guardrails to prevent network transmission and protect sensitive audio sources.
- Recommends using the `whisper` CLI and outlines safe model and performance options.
- Models are downloaded locally on first use.
Metadata
Frequently Asked Questions
What is Openai Whisper Hardened?
Local speech-to-text with the Whisper CLI (no API key). It is an AI Agent Skill for Claude Code / OpenClaw, with 40 downloads so far.
How do I install Openai Whisper Hardened?
Run "/install openai-whisper-hardened" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Openai Whisper Hardened free?
Yes, Openai Whisper Hardened is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Openai Whisper Hardened support?
Openai Whisper Hardened is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Openai Whisper Hardened?
It is built and maintained by Faberlens (@snazar-faberlens); the current version is v1.0.0.
More Skills