Browse categories
Explore
Fiverr Pro
English
$
USD
I will set up claude code with custom agents, hooks, and safe permissions
Devan
About this gig
Most "Claude Code setup" gigs mean copying a CLAUDE.md template. I run this for real: a production system where multiple Claude agents work in isolated git worktrees, respect permission levels, and go through a review gate before anything merges, unattended, for hours at a time.
What I set up is built to survive contact with an actual team: agents that do not step on each other's branches, hooks that stop a destructive command before it runs instead of after, and permissions that match who is allowed to touch what.
What you get:
- A CLAUDE.md and agent configuration tuned to your actual repo, not a generic boilerplate.
- Hooks and guardrails that prevent the failure modes that actually bite teams (force-pushes, stray rm -rf, secrets in commits).
- Documentation so your team understands what was set up and why.
What I need from you: read access to your repo (or a representative sample if it is private/sensitive), a sense of your team's workflow (solo dev vs. multiple engineers vs. CI-integrated), and what tasks you are hoping to hand off to agents.
Get to know Devan
Devan
Daily stack: MCP servers, Claude Code setups, Autonomous agents
- FromCanada
- Member sinceSep 2013
- Avg. response time1 hour
Languages
English
I build infrastructure for AI agents: MCP servers that connect Claude to real tools and data, Claude Code setups for dev teams, and autonomous agents that run workflows unattended.
Coding since 2013, but I only started taking gigs seriously this year.
I run this stack daily on my own multi-agent platform: queued tasks, isolated execution, a review gate before anything merges.
Fast responder, clear about scope before I start, and I will tell you honestly if your project is not a fit.
My Portfolio
FAQ
I already have a basic CLAUDE.md. What does this add?
Most basic setups are a static instructions file. This adds actual guardrails (hooks that block risky commands), scoped permissions (agents cannot touch what they should not), and, in Standard/Premium, a multi-agent workflow instead of one generic assistant doing everything.
Is this safe to run against my production repo?
The setup itself defaults to isolated worktrees so agent work never touches your live branches directly. Changes go through review before merging, same principle as a PR.
Do you need write access to my codebase?
For the setup itself, read access plus permission to add config files is usually enough. If you want me to demonstrate the agent workflow on a real task, we can scope that separately.
What if my team does not use git worktrees today?
That is fine. It is part of what gets set up in the Standard/Premium tiers, with docs explaining the workflow change.
