How to quickly onboard a programmer to a project
many years ago I would worked with https://doxygen.nl/ to document the code but today the easy way will be using Chatgpt Codex or Gemini CLI:
Lets look the differences between the tools:
| Area | Gemini CLI | ChatGPT Codex CLI |
|---|---|---|
| Vendor / model | Google, Gemini 2.5 Pro | OpenAI, GPT-5 / GPT-5-Codex |
| Install | npm i -g @google/gemini-cli or brew install gemini-cli | npm i -g @openai/codex or brew install codex |
| Auth | Google login or Gemini/Vertex API key | Sign in with ChatGPT (Plus/Pro/Team/Enterprise/Edu) or API key |
| License | Apache-2.0, open source | Apache-2.0, open source |
| OS | macOS, Linux, Windows | macOS, Linux; Windows via WSL (experimental) |
| Core behavior | Terminal AI agent with ReAct loop; built-in tools and MCP servers | Local coding agent that reads/edits files and runs commands; MCP support; approval modes |
| Web / grounding | Built-in Google Search grounding and web fetch | Internet access available in Codex Cloud; CLI supports MCP and can delegate |
| Context / quotas | Free tier includes Gemini 2.5 Pro with 1M-token context and ~60 RPM / 1,000 req/day during preview | Uses GPT-5 by default; switchable to GPT-5-Codex; plan-based access via ChatGPT |
| Multimodal | Images/video via Imagen/Veo integrations | Accepts image inputs; IDE and cloud variants can execute and draft PRs |
| Integrations | VS Code (Code Assist agent mode), GitHub Actions, Zed (ACP) | IDE extensions (VS Code, Cursor, Windsurf), Codex Cloud |
Bottom line:
-
Pick Gemini CLI if you want generous free usage, built-in Google Search grounding, and easy MCP-based extensions. blog.google
-
Pick Codex CLI if you want tighter ChatGPT + IDE integration, fine-grained local approvals, and GPT-5-Codex for deep, agentic coding. developers.openai.com
Do it from PowerShell in that folder.
-
Install and sign in
No comments:
Post a Comment