Browse categories
Explore
Fiverr Pro
English
$
USD
I will set up claude code or codex with repo rules and tested workflows
Sage Deutschle
About this gig
Spending more time correcting your coding agent than using it? I configure Claude Code or Codex for your existing codebase and verify one agreed task with your project commands.
BASIC: ONE REPOSITORY
Project rules, build/test commands, one tested workflow and setup notes for one agent on one OS.
STANDARD: TEAM WORKFLOW
Basic plus format/test hooks, up to 2 agreed existing MCP connections and 3 prompt templates.
PREMIUM: CUSTOM TOOL
Standard plus one small custom tool or MCP server with up to 2 agreed operations, tests and a setup guide.
I build developer tooling and use these workflows across Swift, Java, JavaScript and Python projects. You receive configuration files, a change summary, test results and handoff notes.
We agree tool inputs, outputs, access and a passing example before work starts. Existing build failures, missing test suites and custom authentication systems are separate work. Revisions cover the agreed behavior.
Send your OS, agent, stack and task before ordering. You authenticate your own accounts; subscriptions and API usage are your costs. Share redacted errors, never API keys or passwords.
Get to know Sage Deutschle
Sage Deutschle
Minecraft plugins, SwiftUI apps and developer tools
- FromUnited States
- Member sinceJul 2026
Languages
English
I'm Sage, a Computer Science & Engineering student at The Ohio State University. I build Minecraft Paper plugins, fix SwiftUI apps, and configure developer workflows. I created Bayzyl, an open-source Java building toolkit, and co-developed Prismet Arcade for iPhone, iPad and Mac, available on the App Store. My work spans Swift, Java, Linux, servers and automation. Expect an agreed scope, practical updates, tested deliverables and clear setup notes. Message me with your version, goal and relevant error so we can define the right solution.
My Portfolio
FAQ
Why does my coding agent keep ignoring project rules?
I check how project instructions are loaded, remove contradictions, document the actual build and test commands, and try a representative task. The handoff explains what was verified. Clear rules help, but you still need to review generated changes.
Do you need my API keys or subscription?
No. I need read access to the code. I verify on a private copy with my own agent account if you allow it; otherwise I give you a checklist and confirm your run. MCP servers are authenticated by you, with your credentials, never pasted to me.
Does this work with Codex, Cursor or local models?
Claude Code first, and the same AGENTS.md serves the OpenAI Codex CLI. I can wire in local models through Ollama. For Cursor or other tools, ask before ordering and I'll say what carries over.
Will the agent write perfect code afterward?
No one can promise that. A good setup stops repeated mistakes, cuts the permission prompts, and makes auto mode safer, but you should still review what agents change.
Can you work with a private repository?
Yes, once we agree the access, tools and data-handling requirements. Share only code you have permission to provide. External model services and MCP connections are agreed before use; you authenticate your own accounts. Never send production secrets.
