← Back to Skills Marketplace
audio-transcribe
by
TobeyRebecca
· GitHub ↗
· v1.0.0
· MIT-0
81
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install toby-audio-transcribe
Description
Transcribe, diarise, translate, post-process, and structure audio/video with AssemblyAI. Use this skill when the user wants AssemblyAI specifically, needs hi...
Usage Guidance
Do not install or provide credentials until this mismatch is resolved. The skill advertises AssemblyAI but its script points to https://api.heybossai.com and requires SKILLBOSS_API_KEY — meaning your audio and transcripts would be sent to that third-party service, not directly to AssemblyAI. Ask the author to: (1) explain why SKILLBOSS/HeyBoss endpoints are used (is it a proxy/service aggregator?), (2) provide a clear privacy/data-flow statement, and (3) offer an option to use native AssemblyAI endpoints and the official ASSEMBLYAI_API_KEY. If you cannot verify the endpoint and operator, run the skill only in an isolated environment or prefer an alternative skill that explicitly uses assemblyai.com and the official AssemblyAI credential. Also be cautious because the SKILL.md contains long agent-facing instructions (and a detected system-prompt-override pattern), which increases risk if combined with an untrusted remote endpoint.
Capability Analysis
Type: OpenClaw Skill
Name: toby-audio-transcribe
Version: 1.0.0
The skill provides a Node.js CLI for audio transcription via a proxy service (SkillBoss API Hub at api.heybossai.com). The most significant indicator is the unrestricted file system access in scripts/assemblyai.mjs, where the expandHome and writeTextFile functions allow reading from and writing to arbitrary paths (including the user's home directory via '~/') based on agent-provided arguments. This lack of path sanitization or sandboxing represents a high-risk capability that could be exploited for data exfiltration or overwriting sensitive files. Additionally, the SKILL.md metadata falsely attributes the author as 'OpenAI', and there is a discrepancy between the hardcoded proxy URL in the code and the direct AssemblyAI URLs mentioned in the documentation (references/capabilities.md).
Capability Tags
Capability Assessment
Purpose & Capability
Name/description and many assets claim AssemblyAI integration and show AssemblyAI endpoints and env names (ASSEMBLYAI_API_KEY), but the runnable code sets the API base to https://api.heybossai.com/v1 and the declared required environment variable is SKILLBOSS_API_KEY. Requiring SKILLBOSS_API_KEY is not coherent with a skill advertised as 'AssemblyAI' and suggests requests will go to a different service.
Instruction Scope
SKILL.md instructs agents to call AssemblyAI/LLM Gateway endpoints and references ASSEMBLYAI-related environment variables, but the implementation uses the SKILLBOSS API base and SKILLBOSS_API_KEY. The CLI uploads local files and will send audio/transcript data to the base URL configured in the script — currently pointing to a third-party domain rather than assemblyai.com. The skill also exposes raw request passthroughs (e.g., --request/--understanding-request) which is expected for flexibility but increases the risk that arbitrary content could be sent to the configured endpoint.
Install Mechanism
No install spec; it is an instruction-only Node.js CLI and only requires the 'node' binary. That is proportionate for a Node-based CLI skill.
Credentials
The skill requires a single env var named SKILLBOSS_API_KEY (declared as primary). For an AssemblyAI integration one would expect ASSEMBLYAI_API_KEY (or similar) and official AssemblyAI endpoints. Requesting SKILLBOSS_API_KEY is disproportionate given the stated purpose and points to use of a different service for handling audio and transcripts.
Persistence & Privilege
The skill does not request always:true and does not appear to modify other skills' configs. It can be invoked autonomously (default), which is normal; this raises the blast radius only in combination with the other concerns above.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install toby-audio-transcribe - After installation, invoke the skill by name or use
/toby-audio-transcribe - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
AssemblyAI-specific, agent-friendly audio/video transcription and processing.
- Now focused on AssemblyAI features: model routing, diarisation, speaker mapping, and advanced transcript exports (Markdown, JSON, manifest).
- Provides a CLI for transcription, understanding, formatting, LLM-based extraction, and bundled output for agents or downstream workflows.
- Supports language detection/code-switching, translation, and rich output formats (subtitles, paragraphs, sentences).
- Facilitates advanced workflows: speaker name/role mapping, structured extraction, agent-ready outputs.
- Includes model/language lookup commands and local bundled knowledge for easy discovery of supported features.
Metadata
Frequently Asked Questions
What is audio-transcribe?
Transcribe, diarise, translate, post-process, and structure audio/video with AssemblyAI. Use this skill when the user wants AssemblyAI specifically, needs hi... It is an AI Agent Skill for Claude Code / OpenClaw, with 81 downloads so far.
How do I install audio-transcribe?
Run "/install toby-audio-transcribe" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is audio-transcribe free?
Yes, audio-transcribe is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does audio-transcribe support?
audio-transcribe is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created audio-transcribe?
It is built and maintained by TobeyRebecca (@tobeyrebecca); the current version is v1.0.0.
More Skills