Developer fires Gemini for generating empty output instead of code

A developer reports that Google's Gemini AI, after promising to write code, produced only a header line and no actual code. The issue stems from a broken agent pipeline where Gemini attempts to call a code-writing tool but fails to execute. The developer concludes Gemini is unreliable for coding tasks and switches to Claude.