Browse categories
Explore
Fiverr Pro
English
$
USD
I will jarvis ai agent hermes shopify liquid api mcp n8n automation claude code xenbar
Felix
About this gig
Your AI agent installed fine. Then it got your inbox, your store and your server, with no handrail.
I've seen where agent setups really break: permissions far too broad, mcp connectors with unchecked write access, and updates that wreck a working claude code or openclaw hermes setup. I build agents that do the job, and only the job.
What I deliver:
- claude code setup with scoped permissions, hooks and a CLAUDE.md
- claude cowork with limited folders and approval steps, not blanket access
- claude mcp servers, read-only first, with separate write tools
- ai agent builds: jarvis ai and claude ai agent workflows with logging
- hermes agent installs and openclaw to hermes migration
- fix openclaw hermes crashes, exposed gateways and broken updates
- n8n automation and claude automation that alert you when they fail
- shopify automation: shopify liquid fixes, claude code via Shopify's Dev MCP
- api mcp integrations for your CRM and database
I trace each failure to its root, not patch the symptom.
Send me the agent you want running, what it can touch today, and what's breaking. I'll reply with a plan, the risks I see, and what I'd change before you order.
Get to know Felix
Felix
Ai Automation Specialist
- FromUnited States
- Member sinceOct 2026
Languages
English, French
Hi, I'm Felix, an AI automation and agent developer. I build AI agents and automations that save businesses hours of manual work every week.
I specialize in Claude Code and Claude Cowork setups, custom AI agents (Jarvis and Hermes style), MCP integrations, n8n workflows, API connections, and Shopify Liquid customization.
I deliver working, documented solutions with fast replies and clear communication from start to finish.
Your idea, my expertise. Let's bring them to reality.
Other AI Development Services I Offer
FAQ
Is my claude code problem a Basic job, or do I need Standard or Premium?
Basic covers one tool: a claude code install, or one openclaw hermes or n8n automation fix. Standard adds claude cowork or hermes agent setup, claude mcp connections and one working workflow. Premium is the full ai agent stack: jarvis ai, mcp automation, shopify liquid work, hardening .
Do you handle claude cowork and openclaw hermes, or only claude code?
All three. I set up claude code, claude cowork with limited folders and approval steps, and fix openclaw hermes crashes. Moving from OpenClaw to Hermes Agent? I can migrate it, and I check memory and skills after the move. Shopify automation is separate scope, so mention it in your first message.
Do I have to give you my passwords or API keys to get started?
Never your main account password. I need your error output, your OS and what the agent should do. For server work, a temporary limited user or revocable keys is enough, and you can delete them after delivery. I'd also rotate any key we used.
My setup was half-built by someone else. Can you still work with it?
Usually yes. I audit first: what runs, what's exposed, what's broken, and whether patching beats a clean rebuild. You get that verdict before I change anything. If the foundation is shaky I'll say so and point you to Standard or Premium instead of stacking fixes on top.
How long does it take, and what slows it down?
Basic is typically about 2 days, Standard around 4, Premium about a week. Delays come from missing access, unclear goals, or a third-party service that's down or changed its rules. If something blocks me I tell you early, with options. You get a plan first, then a short status note at each stage.
What if it breaks after delivery or an update changes something?
A: Problems tied to my work are covered in your revision rounds, and my handover notes show how to restart, roll back and read logs. Tools like openclaw hermes and claude mcp servers update often, so a later release can break a working setup. If that happens, send the new error and I'll scope it.
Will my ai agent run up a surprise bill while I'm asleep?
It shouldn't with my setup. I add stop conditions to loops, cap parallel subagents, and walk you through spend limits in your Anthropic account. Unattended claude automation on an API key can bill fast, and runaway loops have cost people thousands.
Can I use my Claude Pro or Max plan, or do I need API credits?
For claude code itself, your plan can work. Third-party agents like openclaw hermes are different: Anthropic has blocked plan logins in some tools before, and billing rules for automated use have shifted. I check the current rules and set up the allowed path, usually an API key with a spend limit.
What happens if you can't get my claude mcp server connected?
I tell you plainly what's failing and why. Most cases are config scope, bad paths, stray output on stdout or auth, and I check them in that order. If the cause is outside my control, like a retired endpoint or a provider limit, I show the evidence and the workable alternatives.
Will I own everything, and can I run it without you afterwards?
Yes. It runs in your accounts, on your machine or server, with your keys. Nothing depends on me. Premium includes full handover docs, and every package comes with a short note on restarting and checking it. If you want more later, like n8n automation or shopify automation, we scope separately.
