Use Rendrift media processing with the Copilot CLI via Model Context Protocol.
The Rendrift MCP server enables you to use Rendrift's media processing capabilities directly in the GitHub Copilot CLI. Simply configure your API key, and you'll gain access to 5 tools for processing local media files.
Use cases:
Sign up for Rendrift and grab your free API key (or upgrade to a paid plan for higher limits).
Add the Rendrift MCP server to your Copilot CLI configuration:
~/.copilot/config.json
{
"mcpServers": {
"rendrift": {
"command": "node",
"args": ["/path/to/rendrift/mcp/dist/index.js"],
"env": {
"RENDRIFT_API_KEY": "your-api-key-here",
"RENDRIFT_API_URL": "https://rendrift.io"
}
}
}
}
Test that the server is accessible:
$ copilot -m rendrift:list-tools
rendrift-mcp v0.1.0
Tools:
- probe_media
- generate_thumbnail
- extract_audio
- optimize_image
- generate_waveform
Analyze a media file and return metadata: format, duration, codec info, bitrate, stream information.
file_path (string, required)
Absolute path to the media file.
$ copilot -m rendrift:probe_media \
--file_path="/home/user/video.mp4"
Output: JSON with duration, format, video/audio codecs, bitrate, etc.
Capture a frame from a video/image at a specific timestamp and save it locally. Returns base64 image data (viewable by vision AI).
file_path (string, required) Absolute path to the source media file. time (number, optional, default: 0) Timestamp in seconds to capture frame. format (string, optional, default: webp) Output format: jpeg, webp, or png. width (number, optional) Thumbnail width in pixels (aspect ratio preserved).
$ copilot -m rendrift:generate_thumbnail \
--file_path="/home/user/video.mp4" \
--time=30 \
--format=jpeg \
--width=320
Output: Saves to /home/user/video.thumb.jpeg + returns base64 image data for AI vision analysis.
Extract audio from a video file and save as standalone audio file (mp3, aac, wav, flac).
file_path (string, required) Absolute path to the source media file. format (string, optional, default: mp3) Output format: mp3, aac, wav, or flac. bitrate (string, optional) Audio bitrate (e.g., "128k", "320k").
$ copilot -m rendrift:extract_audio \
--file_path="/home/user/video.mp4" \
--format=mp3 \
--bitrate=192k
Output: Saves to /home/user/video.mp3.
Resize, re-encode, or convert an image to modern formats (WebP, AVIF) to reduce file size.
file_path (string, required) Absolute path to the source image file. format (string, optional, default: webp) Output format: webp, jpeg, png, or avif. width (number, optional) Output width in pixels (aspect ratio preserved). quality (number, optional) Quality level 1-100 (higher = better, larger).
$ copilot -m rendrift:optimize_image \
--file_path="/home/user/photo.jpg" \
--format=webp \
--width=800 \
--quality=85
Output: Saves to /home/user/photo.optimized.webp + returns compression stats.
Analyze audio and return normalized amplitude samples for waveform visualization, plus min/max/avg statistics.
file_path (string, required) Absolute path to audio or video file. samples (number, optional, default: 200) Number of data points to return (max: 10,000).
$ copilot -m rendrift:generate_waveform \
--file_path="/home/user/audio.mp3" \
--samples=500
Output: JSON array of normalized amplitudes (0-1) + statistics (min, max, avg).
Configure the MCP server via environment variables:
RENDRIFT_API_KEY
Your Rendrift API key (required). Get one by signing up at rendrift.io.
RENDRIFT_API_URL
API endpoint (default: https://rendrift.io). Change for staging/testing.
POLL_INTERVAL_MS
Check job status every N milliseconds (default: 1500). Lower = faster feedback, higher = less polling.
POLL_TIMEOUT_MS
Max time to wait for job completion (default: 120000 = 2 min). Increase for large files.
# First, probe the video to get metadata $ copilot -m rendrift:probe_media --file_path="/videos/my_video.mp4" { "duration_seconds": 120, "format": "mp4", "video": { "codec": "h264", "width": 1920, "height": 1080 } } # Generate a thumbnail at 30 seconds $ copilot -m rendrift:generate_thumbnail \ --file_path="/videos/my_video.mp4" \ --time=30 \ --width=320
# Extract audio track $ copilot -m rendrift:extract_audio \ --file_path="/videos/interview.mp4" \ --format=mp3 \ --bitrate=128k # Generates: /videos/interview.mp3
# Use Copilot CLI to process multiple images
for img in *.jpg; do
copilot -m rendrift:optimize_image \
--file_path="$PWD/$img" \
--format=webp \
--width=800 \
--quality=80
done
$ copilot -m rendrift:generate_waveform \
--file_path="/audio/podcast.mp3" \
--samples=1000
# Output: Array of 1000 normalized amplitudes (0-1)
# Save to visualization library or web UI
Sign up for free at rendrift.io. Free tier includes 500 ops/month.
Requests will fail with 429 Too Many Requests. Upgrade your plan or wait until the 1st of next month.
No, the MCP server requires internet connectivity to reach the Rendrift API. For offline processing, see our standalone CLI tools (coming soon).
Output files are saved in the same directory as the input file. For example, /path/to/video.mp4 → /path/to/video.thumb.webp.
Check that your API key is valid, your internet is working, and the API URL is correct. You can test with: curl -H "Authorization: Bearer YOUR_KEY" https://rendrift.io/v1/health
Yes, but ensure you have an appropriate plan quota. Consider upgrading to Pro or Business for production workloads.
© 2026 Rendrift. All rights reserved. · System Status · API Docs