Personal MCP Server

Model Context Protocol (MCP) server providing YouTube transcript fetching and extensible tools for AI assistants.

Quick Start

# Pull and run from GitHub Container Registry (recommended)
docker compose pull
docker compose up -d

# View logs
docker compose logs -f

# Stop
docker compose down

Server runs at http://localhost:8000

Features

• YouTube Transcript Fetching - Multiple formats (plain, structured, SRT, VTT), multi-language support

• Persistent Caching - TTL-based with configurable size and expiration

• Rate Limiting - Prevent abuse (100 requests/minute default)

• Security Hardened - Non-root user, read-only filesystem, minimal privileges

Available Tools

youtube_get_transcript

Fetch YouTube video transcript in various formats.

Parameters:

• video_url_or_id - YouTube URL or video ID

• languages - Preferred languages (default: ["en"])

• format - Output format: plain, structured, srt, vtt (default: plain)

• use_cache - Use cache if available (default: true)

youtube_list_languages

List all available transcript languages for a video.

Parameters:

• video_url_or_id - YouTube URL or video ID

get_server_stats

Get server statistics including cache and rate limit info.

Configuration

Edit docker-compose.yml environment variables:

MCP_SERVER_HOST: 0.0.0.0      # Bind address
MCP_SERVER_PORT: 8000         # Port number

CACHE_ENABLED: true           # Enable caching
CACHE_TTL: 3600               # Cache TTL in seconds
CACHE_MAX_SIZE: 100           # Max cached items

RATE_LIMIT_ENABLED: true      # Enable rate limiting
RATE_LIMIT_REQUESTS: 100      # Requests per window
RATE_LIMIT_WINDOW: 60         # Window in seconds

# Optional YouTube settings
# YOUTUBE_COOKIES: /app/cookies.txt
# YOUTUBE_PROXY_HTTP: http://proxy:8080
# YOUTUBE_PROXY_HTTPS: https://proxy:8080

Development

Build and run locally from source:

docker compose -f docker-compose.build.yml up -d --build

Available Images

From ghcr.io/cloonix/mymcp:

• latest - Latest build from main branch

• main - Main branch

• v1.0.0 - Version tags (when released)

License

MIT