← Back to Skills Marketplace
Transcription
by
AVale-SLAI
· GitHub ↗
· v1.0.4
· MIT-0
110
Downloads
0
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install signalloom-transcription
Description
Transcribe audio and video files using the Signal Loom AI API. Supports MP3, WAV, M4A, MP4, MOV, and more. Runs locally on Apple Silicon for speed and privacy.
Usage Guidance
This skill is inconsistent: it promises local-only transcription but references a remote Signal Loom API and asks for an API key. Before installing, verify whether audio will actually be uploaded (ask the author or inspect the transcribe binary's code). Note that the package does not include the 'transcribe' executable it requires; you must install or trust a third-party binary named 'transcribe'. The included install.sh will modify your ~/.zshrc and send an unauthenticated analytics ping to api.signalloomai.com. If you need strong privacy, avoid installing until the developer provides (1) the transcribe script/binary source, (2) explicit confirmation of local vs remote processing, and (3) a versioned release with matching metadata. If you proceed, run the installer in a controlled environment (sandbox or VM) and monitor network traffic to confirm no unexpected uploads.
Capability Analysis
Type: OpenClaw Skill
Name: signalloom-transcription
Version: 1.0.4
The skill's installation script (install.sh) performs automated telemetry by sending an install ping to an external endpoint (api.signalloomai.com) and modifies the user's shell configuration (.zshrc) to alter the system PATH. While these actions are common in developer tools, they constitute unauthorized environment modification and data signaling. Additionally, there is a discrepancy in SKILL.md which claims audio never leaves the machine while simultaneously requiring a Signal Loom AI API key for processing.
Capability Assessment
Purpose & Capability
The SKILL.md advertises local processing ('audio never leaves your machine') but also repeatedly references the Signal Loom AI API and asks users to set SL_API_KEY; package.json has a signalloom api flag. That contradicts the privacy/locality claim. The skill declares it requires a 'transcribe' binary but does not ship one (no scripts/transcribe present), so it depends on an external binary the installer does not install.
Instruction Scope
Runtime instructions are limited to running a 'transcribe' command on files, which is appropriate, but the included install.sh modifies the user's ~/.zshrc (adds ~/.local/bin to PATH) and fires an unsolicited background analytics POST to api.signalloomai.com. The SKILL.md also instructs users to set SL_API_KEY even though the registry metadata lists no required env vars.
Install Mechanism
There is no network download or third‑party package fetch, which is low risk. However, an install.sh is included even though the skill was listed as instruction-only; the script symlinks into ~/.openclaw/skills, adjusts PATH in ~/.zshrc, and sends a telemetry ping. The transcribe binary referenced by the skill is not installed or included.
Credentials
The registry metadata declares no required env vars, but the README and install output instruct the user to export SL_API_KEY (Signal Loom API key). Asking for an API key would be proportionate if the skill used the remote API, but that conflicts with the advertised 'local' processing. The skill does not request unrelated credentials, but the mismatch between declared and actual env usage is problematic.
Persistence & Privilege
The skill is not forced always-on and does not request elevated privileges. It does persist by symlinking into ~/.openclaw/skills and edits the user's ~/.zshrc (adds PATH), which is typical for simple installers but is a user-impacting change the user should be aware of.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install signalloom-transcription - After installation, invoke the skill by name or use
/signalloom-transcription - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.4
- Maintenance update to install.sh script.
- No user-facing features or documentation changes.
v1.0.3
Version 1.0.3
- Added package.json file for improved dependency and environment management.
v1.0.1
- Added Openclaw-specific metadata including emoji, required binaries, and install instructions.
- Removed mention of direct API key requirement and pricing details from the main description.
- Simplified syntax section and usage examples.
- General refinement and streamlining of documentation for clarity.
v1.0.0
Initial release of the Signal Loom transcription skill:
- Transcribes audio and video files (MP3, WAV, M4A, MP4, MOV, and more) using the Signal Loom AI API.
- Runs locally on Apple Silicon for enhanced speed and privacy.
- Outputs transcripts in JSON, SRT, VTT, and plain text formats.
- Supports up to 100 free transcription minutes per month with a Signalloom API key.
- Includes support for multiple languages and subtitle generation.
Metadata
Frequently Asked Questions
What is Transcription?
Transcribe audio and video files using the Signal Loom AI API. Supports MP3, WAV, M4A, MP4, MOV, and more. Runs locally on Apple Silicon for speed and privacy. It is an AI Agent Skill for Claude Code / OpenClaw, with 110 downloads so far.
How do I install Transcription?
Run "/install signalloom-transcription" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Transcription free?
Yes, Transcription is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Transcription support?
Transcription is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Transcription?
It is built and maintained by AVale-SLAI (@avale-slai); the current version is v1.0.4.
More Skills