I will set up and verify claude code, codex or gemini cli for your codebase


About this gig
Default AI coding agents guess your test commands, touch the wrong files and say "done" before anything runs. I configure Claude Code, OpenAI Codex or Gemini CLI for your codebase and prove the setup works.
What you get:
Custom CLAUDE.md, AGENTS.md or GEMINI.md built from your real code, scripts and rules
Safe permission settings so risky commands cannot run unchecked
Connected MCP servers for code hosting, databases and docs, plus custom skills
Automated checks that run your tests or build after every change
Verification run with real results and a plain-English handover guide
Why me:
I use Claude Code, Codex and Gemini daily and build open-source tools for them: vibe-claude (29 stars) to verify work actually runs before completion, skilladopt to adapt third-party skills to your codebase, and vibe-codex for Codex CLI.
How it works:
100% async in writing with daily updates and no calls. You keep your keys; I never ask for passwords.
What I need: repo access, your test commands and your chosen agent.
Get to know Kyoungsoo Kim
AI Video Producer and AI Automation Specialist
- FromSouth Korea
- Member sinceApr 2022
- Avg. response time1 hour
Languages
English, Korean, Japanese
My Portfolio
Other AI Development Services I Offer
FAQ
Which agent should I choose?
Tell me what you work on and I'll suggest one. Many teams use more than one: I run Claude Code, Codex and Gemini CLI side by side every day, and the same project rules can serve all of them.
Do you need my API keys or account password?
No. You keep your own Anthropic, OpenAI or Google subscription and sign in yourself. I never store client keys or passwords.
What if my project has no tests?
I add a minimal check, such as a build or a smoke run of the program, so the AI's "done" can still be proven.

