Gladia YouTube Transcription (Free)

Transcribe speech from YouTube videos or audio URLs into text using Gladia API with up to 10 free hours of monthly transcription. Use when: you need to summa...

Security Scan

The skill's code, instructions, and required environment variables are consistent with its stated purpose of sending public audio/video URLs to the Gladia transcription API; nothing in the package appears disproportionate or unrelated.

Purpose & Capability

The name/description say 'transcribe YouTube/audio via Gladia'. The only required credential is GLADIA_API_KEY and required binaries are curl and python3 — all directly used by the included script to call Gladia endpoints. There are no unrelated credentials, binaries, or config paths.

Instruction Scope

SKILL.md and the shell script stay within the transcription task: they instruct how to set GLADIA_API_KEY, submit an audio_url to Gladia, poll for results, and save transcripts locally. The instructions do not ask the agent to read unrelated files, system secrets, or transmit data to endpoints other than api.gladia.io. The doc does mention storing keys in ~/.bashrc as an option but also warns against it.

Install Mechanism

There is no install spec and the skill is instruction-only plus a small shell script. Nothing is downloaded or written by an installer; risk from installation is minimal.

Credentials

Only GLADIA_API_KEY is required, which is proportionate for an API-based transcription service. The script only reads this env var. No additional secrets or unrelated API keys are requested.

Persistence & Privilege

The skill is not always-enabled (always:false) and does not request elevated or persistent system privileges. It does not modify other skills or system-wide settings. The normal platform default allowing autonomous invocation remains (disable-model-invocation:false).

Assessment

This skill appears to do exactly what it claims: call the Gladia pre-recorded transcription API for a provided public audio/video URL and save the transcript locally. Before installing: 1) Only provide public URLs (the script and docs already advise against private/unlisted content). 2) Keep your GLADIA_API_KEY secret—prefer session-only env vars or a secrets manager rather than committing it to dotfiles; the SKILL.md warns about this. 3) Be aware of Gladia quota/charges (10 free hours/month) and that transcripts may contain sensitive data. 4) The script writes transcripts into a transcripts/ subdirectory under the skill folder—verify filesystem permissions and cleanup policies if you expect sensitive output. 5) If you do not want the agent to call this skill autonomously, disable autonomous invocation at the agent/platform level.

Video/Audio Transcription Skill

Overview

This skill provides automated transcription for video and audio content using Gladia API. It converts spoken content from YouTube videos, MP3 files, or any accessible audio/video URL into text, which can then be summarized by an LLM.

Use Cases

• YouTube Video Summary - Transcribe YouTube videos for LLM summarization (especially useful for Cantonese/Chinese content without captions)
• MP3/WAV to Text - Convert audio files to transcript
• Video Content Extraction - Extract speech from any publicly accessible video URL
• Podcast Transcription - Convert podcast episodes to text

Service: Gladia API

What is Gladia?

Gladia is an audio transcription API that supports multiple languages including Cantonese. It provides both async (pre-recorded) and real-time transcription.

Free Tier (as of March 2026)

Feature	Free Tier
Monthly transcription	10 hours
Renewal	Monthly (resets automatically)
New streams limit	5 per minute
Languages	All included
Cost after quota	$0.61/hour (async) / $0.75/hour (real-time)

How to Sign Up

1. Visit gladia.io
2. Click "Try for free" → Sign up with email
3. Go to Dashboard → API Keys
4. Create a new API key
5. Copy the key (format: xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx)

Checking Usage

1. Log in to Gladia Dashboard
2. Navigate to Usage or Billing section
3. View current month consumption (hours/minutes used)

Alternatively, you can check via API:

curl -X GET "https://api.gladia.io/v2/usage" -H "x-gladia-key: YOUR_API_KEY"

Setup

Step 1: Save Your API Key

Recommended: Set in current session

export GLADIA_API_KEY="your-api-key-here"

Or add to ~/.bashrc (ensure ~/.bashrc is in .gitignore):

echo 'export GLADIA_API_KEY="your-api-key-here"' >> ~/.bashrc
source ~/.bashrc

Note: Storing secrets in shell rc files is discouraged due to risk of accidental commits. Prefer setting the environment variable directly in your session or use a secrets manager.

Step 2: Verify Key Works

Test that your API key is valid by checking usage:

curl -X GET "https://api.gladia.io/v2/usage" -H "x-gladia-key: $GLADIA_API_KEY"

If successful, you'll see your usage information. If you get an auth error, check your API key.

How to Use

Command Line

# Navigate to skill directory (where you installed the skill)
cd /path/to/video-transcription

# Basic usage
./scripts/youtube_transcribe.sh "YOUTUBE_URL"

# Save to specific file
./scripts/youtube_transcribe.sh "YOUTUBE_URL" /path/to/output.txt

Via OpenClaw

1. Provide a YouTube URL or video link
2. The skill will:
   • Submit transcription job to Gladia
   • Poll for completion (~1-2 min for 10-15 min videos)
   • Return the full transcript

Script Location

/path/to/video-transcription/scripts/youtube_transcribe.sh

Configuration

Environment Variables

Variable	Required	Description
GLADIA_API_KEY	Yes	Your Gladia API key

Output

The script saves transcripts to a transcripts/ subdirectory in the skill folder:

/path/to/video-transcription/transcripts/

Privacy & Security Notes

• NEVER share your API key publicly
• NEVER include your API key in any skill documentation or code commits
• IMPORTANT: Do NOT store API keys in shell rc files (~/.bashrc, ~/.zshrc) or config files that might be committed to version control
• Use session-only environment variables: export GLADIA_API_KEY='your-key'
• Or use a secrets manager (e.g., 1Password, AWS Secrets Manager)
• Output transcripts may contain sensitive content - handle accordingly

Limitations

• Video must be publicly accessible (no private/unlisted content)
• Audio quality affects transcription accuracy
• Some copyrighted content may have restrictions
• Processing time depends on video length (~10 seconds per minute of video)
• Free quota resets monthly; excess usage incurs charges

Troubleshooting

"Failed to call the url"

• The video URL may be inaccessible or private
• Try a different video URL

"Quota exceeded"

• You've reached the 10-hour monthly limit
• Wait for quota reset next month, or upgrade to paid plan

"Authentication failed"

• Check your API key is correct
• Ensure GLADIA_API_KEY environment variable is set

Alternative Services

If Gladia quota is exhausted:

Service	Free Tier	Notes
AssemblyAI	Limited	Requires credit card
Deepgram	$0 credit	Pay-per-use
YouTube Transcript	Free (if available)	Only works if video has captions

Future Enhancements

Potential improvements:

• Add speaker diarization (identify different speakers)
• Support real-time transcription
• Automatic LLM summarization after transcription
• Multi-language translation
• Save transcripts to cloud storage

---

Last updated: March 2026