Deployment

Deployment Options

erpnext-agent exposes its MCP server (console script erpnext-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. Replace the <your-…> placeholders with the values from the Configuration / Environment Variables 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": {
    "erpnext-mcp": {
      "command": "uvx",
      "args": ["--from", "erpnext-agent", "erpnext-mcp"],
      "env": {
        "ERPNEXT_URL": "<your-erpnext_url>",
        "ERPNEXT_TOKEN": "<your-erpnext_token>"
      }
    }
  }
}

2. Streamable-HTTP (local process)

Run the server as a long-lived HTTP process:

uvx --from erpnext-agent erpnext-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": {
    "erpnext-mcp": {
      "command": "uvx",
      "args": ["--from", "erpnext-agent", "erpnext-mcp", "--transport", "streamable-http", "--port", "8000"],
      "env": {
        "TRANSPORT": "streamable-http",
        "HOST": "0.0.0.0",
        "PORT": "8000",
        "ERPNEXT_URL": "<your-erpnext_url>",
        "ERPNEXT_TOKEN": "<your-erpnext_token>"
      }
    }
  }
}

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

{
  "mcpServers": {
    "erpnext-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": {
    "erpnext-mcp": {
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "TRANSPORT=stdio",
        "-e", "ERPNEXT_URL=<your-erpnext_url>",
        "-e", "ERPNEXT_TOKEN=<your-erpnext_token>",
        "knucklessg1/erpnext-agent:latest"
      ]
    }
  }
}

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

docker run -d --name erpnext-mcp -p 8000:8000 \
  -e TRANSPORT=streamable-http \
  -e PORT=8000 \
  -e ERPNEXT_URL="<your-erpnext_url>" \
  -e ERPNEXT_TOKEN="<your-erpnext_token>" \
  knucklessg1/erpnext-agent:latest
# or, from a clone of this repo:
docker compose -f docker/mcp.compose.yml up -d

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

(c) From a local checkout with uv:

uv run erpnext-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": {
    "erpnext-mcp": { "url": "http://erpnext-mcp.arpa/mcp" }
  }
}

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

This page covers running erpnext-agent 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. To provision the ERPNext / Frappe site it connects to, see Backing Platform.

erpnext-agent ships two console scripts: an MCP server (erpnext-mcp) that exposes a typed, deterministic tool surface a policy router / agent calls, and an optional A2A agent server (erpnext-agent) that wraps those tools in a Pydantic-AI agent. Most deployments run the MCP server; the agent server is documented in Agent server below.

Run the MCP server

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

stdio (default)streamable-httpsse

erpnext-mcp

For IDE / desktop MCP clients that launch the server as a subprocess.

erpnext-mcp --transport streamable-http --host 0.0.0.0 --port 8000

A network server with a /health endpoint and /mcp route.

erpnext-mcp --transport sse --host 0.0.0.0 --port 8000

Health check (HTTP transports):

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

Configuration (environment)

erpnext-agent is configured entirely from the environment. The required set:

Var	Default	Meaning
ERPNEXT_URL	http://localhost:8000	ERPNext / Frappe site base URL
ERPNEXT_TOKEN	(empty)	API token in api_key:api_secret form

Optional connection and runtime settings:

Var	Default	Meaning
ERPNEXT_AGENT_USERNAME	(empty)	Username for session login (alternative to a token)
ERPNEXT_AGENT_PASSWORD	(empty)	Password for session login
ERPNEXT_AGENT_SSL_VERIFY	True	Verify TLS (set False for self-signed homelab certs)
AUTHENTICATIONTOOL	True	Register the authentication tool set
RESOURCETOOL	True	Register the resource (document/method) tool set
HOST / PORT / TRANSPORT	0.0.0.0 / 8000 / stdio	HTTP transport binding

The connector remains inactive when no ERPNEXT_URL / credentials are supplied. A template lives in .env.example — copy it to .env and fill in your site endpoint and token.

Docker Compose

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

services:
  erpnext-agent-mcp:
    image: knucklessg1/erpnext-agent:latest
    container_name: erpnext-agent-mcp
    hostname: erpnext-agent-mcp
    restart: always
    env_file:
      - ../.env
    environment:
      - PYTHONUNBUFFERED=1
      - HOST=0.0.0.0
      - PORT=8000
      - TRANSPORT=streamable-http
      - ERPNEXT_URL
      - ERPNEXT_TOKEN
    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 ERPNEXT_URL / ERPNEXT_TOKEN
docker compose -f docker/mcp.compose.yml up -d
docker compose -f docker/mcp.compose.yml logs -f

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
erpnext-agent.arpa {
    tls internal
    reverse_proxy erpnext-agent-mcp:8000
}

# Public — automatic Let's Encrypt
erpnext-agent.example.com {
    reverse_proxy erpnext-agent-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=erpnext-agent.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 erpnext-agent.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": {
    "erpnext-agent": {
      "command": "uv",
      "args": ["run", "erpnext-mcp"],
      "env": {
        "ERPNEXT_URL": "https://your-erpnext:8000",
        "ERPNEXT_TOKEN": "your_api_key:your_api_secret"
      }
    }
  }
}

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

Agent server

In addition to the MCP server, erpnext-agent ships a Pydantic-AI agent server (console script erpnext-agent). It connects to one or more MCP servers and exposes a conversational agent over A2A:

# Run the agent server, wiring it to a running MCP endpoint
erpnext-agent --host 0.0.0.0 --port 8001 --mcp-url http://erpnext-agent-mcp:8000/mcp

The agent reads its toolset from an mcp_config.json (default: the bundled erpnext_agent/mcp_config.json) or from the --mcp-url of an already-running MCP server. A companion Compose service mirrors the MCP stack — publish the agent on its own port (for example :8001) and set MCP_URL to the MCP server it should call:

services:
  erpnext-agent:
    image: knucklessg1/erpnext-agent:latest
    container_name: erpnext-agent
    hostname: erpnext-agent
    restart: always
    entrypoint: ["erpnext-agent"]
    env_file:
      - ../.env
    environment:
      - PYTHONUNBUFFERED=1
      - HOST=0.0.0.0
      - PORT=8001
      - MCP_URL=http://erpnext-agent-mcp:8000/mcp
    ports:
      - "8001:8001"
    depends_on:
      - erpnext-agent-mcp

Provision an LLM provider for the agent through the standard agent-utilities environment variables (provider/model selection via --provider / --model-id).