Skip to content

Deployment

Deployment Options

media-downloader exposes its MCP server (console script media-downloader-mcp) four ways. Pick the row that matches where the server runs relative to your MCP client, then copy the matching mcp_config.json below. Add the service-connection environment variables documented in the Configuration section.

# Option Transport Where it runs mcp_config.json key
1 stdio stdio client launches a subprocess command
2 Streamable-HTTP (local) streamable-http a local network port command or url
3 Local container / uv stdio or streamable-http Docker / Podman / uv on this host command or url
4 Remote URL streamable-http a remote host behind Caddy url

1. stdio (local subprocess)

The client launches the server over stdio via uvx — best for local IDEs (Cursor, Claude Desktop, VS Code):

{
  "mcpServers": {
    "media-downloader-mcp": {
      "command": "uvx",
      "args": ["--from", "media-downloader", "media-downloader-mcp"]
    }
  }
}

2. Streamable-HTTP (local process)

Run the server as a long-lived HTTP process:

uvx --from media-downloader media-downloader-mcp --transport streamable-http --host 0.0.0.0 --port 8000
curl -s http://localhost:8000/health        # {"status":"OK"}

Then either let the client launch it:

{
  "mcpServers": {
    "media-downloader-mcp": {
      "command": "uvx",
      "args": ["--from", "media-downloader", "media-downloader-mcp", "--transport", "streamable-http", "--port", "8000"],
      "env": {
        "TRANSPORT": "streamable-http",
        "HOST": "0.0.0.0",
        "PORT": "8000"
      }
    }
  }
}

…or connect to the already-running process by URL:

{
  "mcpServers": {
    "media-downloader-mcp": { "url": "http://localhost:8000/mcp" }
  }
}

3. Local container / uv

(a) Launch a container directly from mcp_config.json (stdio over the container — no ports to manage). Swap docker for podman for a daemonless runtime:

{
  "mcpServers": {
    "media-downloader-mcp": {
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "TRANSPORT=stdio",
        "knucklessg1/media-downloader:latest"
      ]
    }
  }
}

(b) Run a local streamable-http container, then connect by URL:

docker run -d --name media-downloader-mcp -p 8000:8000 \
  -e TRANSPORT=streamable-http \
  -e PORT=8000 \
  knucklessg1/media-downloader:latest
# or, from a clone of this repo:
docker compose -f docker/mcp.compose.yml up -d
{
  "mcpServers": {
    "media-downloader-mcp": { "url": "http://localhost:8000/mcp" }
  }
}

(c) From a local checkout with uv:

uv run media-downloader-mcp --transport streamable-http --port 8000

4. Remote URL (deployed behind Caddy)

When the server is deployed remotely (e.g. as a Docker service) and published through Caddy on the internal *.arpa zone, connect with the "url" key — no local process or image required:

{
  "mcpServers": {
    "media-downloader-mcp": { "url": "http://media-downloader-mcp.arpa/mcp" }
  }
}

Caddy reverse-proxies http://media-downloader-mcp.arpa to the container's :8000 streamable-http listener; http://media-downloader-mcp.arpa/health returns {"status":"OK"} when the service is live.

This page covers running media-downloader as a long-lived server: the transports, a Docker Compose stack, putting it behind a Caddy reverse proxy, and giving it a DNS name with Technitium. It also documents the optional A2A agent server.

media-downloader ships an MCP server (console script media-downloader-mcp) and a Pydantic-AI agent server (console script media-downloader-agent). The agent connects to the MCP server over MCP_URL and exposes a conversational A2A interface with an optional web UI.

Run the MCP server

The transport is selected with --transport (or the TRANSPORT env var):

media-downloader-mcp
For IDE / desktop MCP clients that launch the server as a subprocess.

media-downloader-mcp --transport streamable-http --host 0.0.0.0 --port 8000
A network server with a /health endpoint and /mcp route.

media-downloader-mcp --transport sse --host 0.0.0.0 --port 8000

Health check (HTTP transports):

curl -s http://localhost:8000/health        # {"status":"OK"}

Configuration (environment)

media-downloader is configured entirely from the environment. The required set is small — the downloader needs no external service. Copy .env.example to .env and fill in only what you use:

Var Default Meaning
HOST 0.0.0.0 Bind address for HTTP transports
PORT 8000 Listen port for HTTP transports
TRANSPORT stdio stdio, streamable-http, or sse
YT_DLP_PATH /usr/bin/yt-dlp Path to the yt-dlp binary
BREW_INSTALL_CMD brew install yt-dlp Install hint surfaced for macOS hosts
ENABLE_OTEL True Emit OpenTelemetry traces
EUNOMIA_TYPE none Authorization mode: none, embedded, or remote
EUNOMIA_POLICY_FILE mcp_policies.json Local policy file for embedded mode

Optional telemetry (OTEL_EXPORTER_OTLP_*) and Eunomia remote (EUNOMIA_REMOTE_URL) variables are documented inline in .env.example; each remains inactive when its credentials are absent.

Docker Compose

The repo ships docker/mcp.compose.yml. It reads a sibling .env and publishes the HTTP server on :8000:

services:
  media-downloader-mcp:
    image: knucklessg1/media-downloader:latest
    container_name: media-downloader-mcp
    hostname: media-downloader-mcp
    restart: always
    env_file:
      - ../.env
    environment:
      - PYTHONUNBUFFERED=1
      - HOST=0.0.0.0
      - PORT=8000
      - TRANSPORT=streamable-http
    ports:
      - "8000:8000"
    healthcheck:
      test: ["CMD", "python3", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/health')"]
      interval: 30s
      timeout: 10s
      retries: 3
cp .env.example .env          # then edit values
docker compose -f docker/mcp.compose.yml up -d
docker compose -f docker/mcp.compose.yml logs -f

Agent server (A2A)

media-downloader also ships a Pydantic-AI agent. The console script media-downloader-agent connects to the MCP server over MCP_URL and serves a conversational A2A endpoint (with an optional web UI) on its own port:

export MCP_URL=http://localhost:8000/mcp
media-downloader-agent --provider openai --model-id gpt-4o --host 0.0.0.0 --port 9000

The repo ships docker/agent.compose.yml, which provisions the MCP server and the agent together. The agent waits on the MCP server and is wired to it by container name:

services:
  media-downloader-mcp:
    image: knucklessg1/media-downloader:latest
    hostname: media-downloader-mcp
    environment:
      - HOST=0.0.0.0
      - PORT=8000
      - TRANSPORT=streamable-http
    ports: ["8000:8000"]

  media-downloader-agent:
    image: knucklessg1/media-downloader:latest
    depends_on: [media-downloader-mcp]
    command: ["media-downloader-agent"]
    environment:
      - HOST=0.0.0.0
      - PORT=9000
      - MCP_URL=http://media-downloader-mcp:8000/mcp
      - PROVIDER=${PROVIDER:-openai}
      - MODEL_ID=${MODEL_ID:-gpt-4o}
      - ENABLE_WEB_UI=True
    ports: ["9000:9000"]
docker compose -f docker/agent.compose.yml up -d
Service Port Role
media-downloader-mcp 8000 MCP tool surface (/mcp, /health)
media-downloader-agent 9000 A2A agent + web UI, wired via MCP_URL

Behind a Caddy reverse proxy

Expose the HTTP server on a hostname with automatic TLS. Add to your Caddyfile:

# Internal (self-signed) — homelab .arpa zone
media-downloader.arpa {
    tls internal
    reverse_proxy media-downloader-mcp:8000
}
# Public — automatic Let's Encrypt
media-downloader.example.com {
    reverse_proxy media-downloader-mcp:8000
}

Reload Caddy:

docker compose -f services/caddy/compose.yml exec caddy caddy reload --config /etc/caddy/Caddyfile

DNS with Technitium

Point the hostname at the host running Caddy. Via the Technitium API:

curl -s "http://technitium.arpa:5380/api/zones/records/add" \
  --data-urlencode "token=$TECHNITIUM_DNS_TOKEN" \
  --data-urlencode "domain=media-downloader.arpa" \
  --data-urlencode "zone=arpa" \
  --data-urlencode "type=A" \
  --data-urlencode "ipAddress=10.0.0.10" \
  --data-urlencode "ttl=3600"

…or add an A record media-downloader.arpa → <caddy-host-ip> in the Technitium web console (http://technitium.arpa:5380). The ecosystem technitium-dns-mcp automates this as a tool.

Register with an MCP client

Add to your client's mcp_config.json:

{
  "mcpServers": {
    "media-downloader": {
      "command": "uv",
      "args": ["run", "media-downloader-mcp"],
      "env": {
        "DOWNLOAD_DIRECTORY": "/data/downloads",
        "AUDIO_ONLY": "False"
      }
    }
  }
}

For a remote HTTP server, point the client at http://media-downloader.arpa/mcp instead.