From Zero To Production

A complete end-to-end guide: install mcp2cli, connect to MCP servers, configure for production, and set up monitoring.

Phase 1: Install & Explore

Install

# From source
git clone https://github.com/mcp2cli/source-code.git
cd source-code
cargo install --path .

# Verify
mcp2cli --version

Try the Demo

No server needed — the built-in demo mode lets you explore:

mcp2cli config init --name demo --app bridge \
  --transport streamable_http --endpoint https://demo.invalid/mcp

mcp2cli use demo

# Discover capabilities
mcp2cli ls

# Call tools
mcp2cli echo --message "hello world"
mcp2cli add --a 5 --b 3

# Read resources
mcp2cli get demo://resource/readme.md

# Health check
mcp2cli doctor

Phase 2: Connect to Your Server

For HTTP Servers

mcp2cli config init --name myserver --app bridge \
  --transport streamable_http \
  --endpoint http://your-server:3001/mcp

mcp2cli use myserver
mcp2cli doctor           # Verify connection
mcp2cli ls               # Discover capabilities

For Stdio Servers

mcp2cli config init --name myserver --app bridge \
  --transport stdio \
  --stdio-command your-server-binary \
  --stdio-args '--config=config.yaml'

mcp2cli use myserver
mcp2cli doctor
mcp2cli ls

Quick Test Without Config

# Just point and go
mcp2cli --url http://your-server:3001/mcp ls
mcp2cli --stdio "your-server-binary" ls

Phase 3: Configure for Your Team

Create a Production Config

schema_version: 1

app:
  profile: bridge

server:
  display_name: Production API Server
  transport: streamable_http
  endpoint: https://mcp-api.company.com/mcp

defaults:
  output: human
  timeout_seconds: 120

logging:
  level: warn
  format: pretty
  outputs:
    - kind: stderr

auth:
  browser_open_command: xdg-open

events:
  enable_stdio_events: true
  http_endpoint: "http://monitoring.internal:9090/mcp-events"

profile:
  display_name: "Production CLI"
  aliases:
    long-running-operation: lro
  hide:
    - debug-internal
    - test-tool
  groups:
    data:
      - query
      - export
      - import
    admin:
      - user-create
      - user-delete
      - config-update
  flags:
    query:
      database: db
      collection: col
  resource_verb: fetch

Set Up Aliases

mcp2cli link create --name prod
mcp2cli link create --name prod --dir /usr/local/bin    # System-wide

# Now use it
prod ls
prod doctor
prod query --db users --col profiles

Multi-Environment Setup

# Development
mcp2cli config init --name dev --transport stdio \
  --stdio-command ./target/debug/my-server

# Staging
mcp2cli config init --name staging --transport streamable_http \
  --endpoint https://staging-mcp.company.com/mcp

# Production
# (use the YAML above)

# Create aliases for each
mcp2cli link create --name dev
mcp2cli link create --name staging
mcp2cli link create --name prod

Phase 4: Authentication

# Login (stores token locally)
prod auth login
# Enter token: <paste-your-token>

# Verify
prod auth status
# → Auth state: active

# All subsequent commands use the stored token
prod ls
prod query --db users

Phase 5: Optimize Performance

Enable Daemon Mode

For stdio servers or frequent usage:

# Start daemon (initializes connection once)
mcp2cli daemon start prod

# All commands now route through the daemon (~50ms instead of ~2s)
prod ls
prod echo --message test
prod query --db users

# Check daemon status
mcp2cli daemon status prod

# Stop when done
mcp2cli daemon stop prod

Tune Timeouts

defaults:
  timeout_seconds: 120     # 2 min for normal operations

# For slow operations, override per-command:
# prod --timeout 600 export --dataset full
# prod --timeout 0 long-running-import

Phase 6: CI/CD Integration

GitHub Actions

name: Deploy via MCP
on:
  push:
    branches: [main]

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Install mcp2cli
        run: cargo install --path .

      - name: Configure
        run: |
          mcp2cli config init --name prod --transport streamable_http \
            --endpoint "${{ secrets.MCP_ENDPOINT }}"
          mcp2cli use prod

      - name: Authenticate
        run: |
          echo "${{ secrets.MCP_TOKEN }}" | mcp2cli auth login

      - name: Pre-deploy health check
        run: prod --timeout 10 doctor

      - name: Deploy
        run: |
          prod --json deploy \
            --version "${{ github.sha }}" \
            --environment production \
            --timeout 300

      - name: Post-deploy verification
        run: |
          prod --timeout 10 ping
          HEALTH=$(prod --json doctor | jq -r '.summary')
          echo "Post-deploy health: $HEALTH"

Shell Script Pipeline

#!/bin/bash
set -euo pipefail

# Pre-flight
prod --timeout 10 doctor || { echo "Server unhealthy"; exit 1; }

# Deploy with background job
RESULT=$(prod --json deploy --version "$VERSION" --background)
JOB_ID=$(echo "$RESULT" | jq -r '.data.job_id')

# Wait
timeout 600 prod jobs wait "$JOB_ID" || {
  prod jobs cancel "$JOB_ID"
  exit 1
}

# Verify
prod --timeout 5 ping
echo "Deploy complete"

Phase 7: Monitoring & Observability

Event Sinks

events:
  enable_stdio_events: true

  # Send events to monitoring system
  http_endpoint: "http://monitoring.internal:9090/mcp-events"

  # Stream to local dashboard
  sse_endpoint: "127.0.0.1:9091"

  # Log to file via command
  command: |
    echo "$(date -Iseconds) [${MCP_EVENT_TYPE}] ${MCP_EVENT_MESSAGE}" \
      >> /var/log/mcp2cli/events.log

Health Check Probe

#!/bin/bash
# health-probe.sh — run every 30s from cron/systemd

if prod --timeout 5 --json ping >/dev/null 2>&1; then
  echo "mcp_server_up 1" | curl -s --data-binary @- http://pushgateway:9091/metrics/job/mcp
else
  echo "mcp_server_up 0" | curl -s --data-binary @- http://pushgateway:9091/metrics/job/mcp
  # Alert
  curl -s -X POST "$SLACK_WEBHOOK" -d '{"text": "🔴 MCP server down"}'
fi

Structured Logging

logging:
  level: info                          # info for production visibility
  format: json                         # JSON for log aggregation
  outputs:
    - kind: stderr
    - kind: file
      path: /var/log/mcp2cli/mcp2cli.log

Phase 8: Team Onboarding

Shared Config Repository

# Check configs into your repo
mkdir -p mcp-configs
cp ~/.local/share/mcp2cli/configs/*.yaml mcp-configs/

# Team members use:
export MCP2CLI_CONFIG_DIR=./mcp-configs
mcp2cli config list

Profile Overlays for Usability

Create user-friendly command surfaces:

profile:
  display_name: "Acme Platform CLI"
  aliases:
    execute-sql-query: query           # Short aliases
    create-deployment: deploy
    get-application-status: status
  groups:
    data:                              # Logical grouping
      - query
      - export
      - import
    deploy:
      - deploy
      - rollback
      - status

Documentation for Your Team

# Show your team what's available
prod ls
prod --help
prod data --help
prod deploy --help

Production Checklist

Item	Command	Status
Config created	`mcp2cli config show --name prod`	☐
Connection verified	`prod doctor`	☐
Auth configured	`prod auth status`	☐
Aliases created	`prod ls`	☐
Daemon running	`mcp2cli daemon status prod`	☐
Timeouts configured	Check `defaults.timeout_seconds`	☐
Events configured	Check `events` section	☐
Health probe active	Cron/systemd probe running	☐
CI/CD integration	Pipeline deploys with `prod`	☐
Team access	Shared config repo	☐
Profile customized	Commands are intuitive	☐