clawdbot-workspace/proposals/2026-02-17-claude-code-mcp-n8n-coach.md
2026-02-17 23:03:48 -05:00

6.2 KiB

Upwork Application: Claude Code + MCP + n8n Automation Coach

Job URL: https://www.upwork.com/jobs/~022023795669666268090 Client: Virginia Beach, $216K spent, 5.0 stars (117 reviews), Sales & Marketing, mid-sized company Connect cost: 31 Connects (60 available) Bid range: High $102 | Avg $48.68 | Low $15 Posted: 4 hours ago | Last viewed: 3 hours ago | Interviewing: 3 | Proposals: 20-50 Rate: $85/hr


Cover Letter

Hey -- I use Claude Code + MCP + n8n literally every single day. This is my actual daily workflow, not something I learned for a gig.

Here's what I bring to the table:

-- I run Claude Code as my primary dev environment on Mac (M-series). I've configured it with custom system prompts, project-level CLAUDE.md files, MCP tool integrations, and multi-agent workflows where Claude Code spawns sub-agents for parallel work.

-- I've built and deployed 30+ MCP servers -- filesystem, GitHub, Brave Search, PostgreSQL/Supabase, Google Workspace, Slack, custom REST API wrappers, and more. I understand the protocol at a deep level (JSON-RPC, stdio vs SSE transport, tool schemas).

-- My n8n instance (self-hosted) runs production workflows including: Gmail monitoring -> AI scoring -> auto-actions, CRM pipelines with webhook triggers, multi-step content generation chains, and Slack/Discord notification routing.

-- I've connected all three into a unified stack: Claude Code calls MCP tools that trigger n8n workflows, which feed back into AI pipelines. It's the "agent/bot system" you're describing -- I've already built it for myself.

For the 2-hour session, here's my proposed agenda:

HOUR 1 -- Foundation & Quick Wins

  • Audit your current Mac dev setup (Homebrew, Node, Python, shell)
  • Install & configure Claude Code with your Anthropic API key
  • Set up your first project with a proper CLAUDE.md context file
  • Install 3-4 essential MCP servers (filesystem, GitHub, Brave search)
  • Live demo: use Claude Code to build something real in your codebase
  • Validate the full loop works end-to-end

HOUR 2 -- n8n + Architecture

  • Set up n8n (local Docker or hosted -- I'll recommend based on your needs)
  • Build a working automation: e.g., email trigger -> AI processing -> Slack notification
  • Map out your company's automation architecture (what connects to what)
  • Document the full stack so you can reference it after the call
  • Prioritized roadmap: what to build next, in what order

I'll also send you a pre-session prep checklist so we hit the ground running with zero wasted time.

What makes me different from the other applicants: I'm not just familiar with these tools -- I've pushed them to their limits. I run a personal AI assistant (Buba) that orchestrates Claude Code, MCP servers, n8n workflows, Discord bots, iMessage integrations, and more. I know where the sharp edges are and how to work around them.

Portfolio: https://portfolio.mcpengage.com


Screening Question Answers

Q1: Have you personally set up Claude Code locally before? Describe the setup.

Yes -- Claude Code is my primary development environment. My setup:

  • Mac M-series, Claude Code installed via npm (globally)
  • Custom system prompts and project-level CLAUDE.md files for per-project context
  • MCP server configs in ~/.claude/settings.json defining tool integrations
  • I use extended thinking mode for complex architecture decisions
  • Multi-agent workflows where Claude Code spawns parallel sub-agents for heavy builds
  • I've configured it across multiple projects -- full-stack web apps, CLI tools, API integrations, browser automation

I also run an autonomous AI assistant built on top of Claude that uses Claude Code as one of its capabilities. I've logged thousands of hours in this environment.

Q2: Have you set up MCP servers? Which ones?

I've built and deployed 30+ MCP servers:

Production servers I run daily:

  • Filesystem (read/write project files)
  • GitHub (issues, PRs, CI runs, API queries)
  • Brave Search (web research)
  • PostgreSQL / Supabase (database operations)
  • Google Workspace (Gmail, Calendar, Drive, Sheets)
  • Slack + Discord integrations
  • Custom REST API wrappers for various SaaS tools

I've also built custom MCP servers from scratch using the TypeScript SDK, including servers that bridge to n8n workflows. I understand the protocol deeply -- JSON-RPC 2.0, stdio vs SSE transport, tool/resource/prompt primitives, and how to debug connection issues.

Q3: Share 1-2 examples of n8n workflows you've built (briefly).

  1. Email Intelligence Pipeline: Gmail trigger monitors incoming emails -> AI node scores and categorizes each email -> conditional routing (high-priority gets Slack alert + auto-drafted response, medium gets queued, low gets archived) -> results logged to Google Sheets for weekly review. Runs 24/7, processes ~50 emails/day.

  2. CRM Lead Scoring & Outreach: Webhook receives new leads from multiple sources -> enrichment via API calls (company data, social profiles) -> AI scoring model rates lead quality -> qualified leads get auto-added to CRM pipeline with personalized outreach draft -> team gets Slack notification with lead summary. Connected to GoHighLevel CRM.

Q4: If we do a 2-hour call, what is your plan/agenda?

HOUR 1 -- Foundation & Quick Wins (first 60 min)

  • 0:00 - Quick audit of your Mac setup (5 min)
  • 0:05 - Install & configure Claude Code (15 min)
  • 0:20 - Set up CLAUDE.md project context (10 min)
  • 0:30 - Install 3-4 MCP servers (filesystem, GitHub, search) (15 min)
  • 0:45 - Live build: use Claude Code on a real task from YOUR codebase (15 min)

HOUR 2 -- n8n + Architecture (second 60 min)

  • 1:00 - Set up n8n (Docker local or hosted) (15 min)
  • 1:15 - Build a working workflow: trigger -> AI -> action (20 min)
  • 1:35 - Architecture mapping: your company's automation stack (15 min)
  • 1:50 - Documentation + prioritized next steps roadmap (10 min)

You'll leave with: working Claude Code, 3+ MCP servers, 1 n8n workflow, and a written roadmap for the larger implementation.

Q5: What is your hourly rate?

$85/hr. For context -- I bring production experience with this exact stack (not theoretical knowledge), and I can compress weeks of trial-and-error into a focused 2-hour session. I'm also available for the ongoing consulting and larger implementation project you mentioned.