Gemini CLI Chatbot Documentation

Gemini CLI Architecture Overview

Gemini CLI is an open-source terminal interface for Google's Gemini models built entirely using POSIX Shell / Bash, curl, and jq. It requires no Python or Node.js runtime, ensuring sub-second startup times and less than 1 MB of memory usage.

API Key Setup & Verification

Obtain a Google Gemini API Key from Google AI Studio and configure your shell environment:

# Temporary session export
export GEMINI_API_KEY="AIzaSyYourSecretKeyHere"

# Permanent shell export (add to ~/.bashrc or ~/.zshrc)
echo 'export GEMINI_API_KEY="AIzaSyYourSecretKeyHere"' >> ~/.bashrc
source ~/.bashrc

# Test connectivity
./gemini.sh "Hello Gemini, are you online?"

Pure Bash SSE Streaming Parser

The script uses curl --no-buffer -N -s connected directly to a while-read stream parser:

# Streaming line processor
curl -s -N -X POST "$API_ENDPOINT" \
  -H "Content-Type: application/json" \
  -d "$PAYLOAD" | while IFS= read -r line; do
    if [[ "$line" =~ ^data:\ (.*) ]]; then
      token=$(echo "${BASH_REMATCH[1]}" | jq -r '.candidates[0].content.parts[0].text // empty')
      printf "%s" "$token"
    fi
done

Interactive REPL Commands

Unix Piping Workflows

Automate terminal diagnostics by piping standard output into Gemini CLI:

# 1. Analyze git diff before committing
git diff | gemini "Review this diff and suggest a conventional commit message"

# 2. Diagnose failing systemd service
journalctl -u nginx.service -n 50 --no-pager | gemini "Explain the root cause of this failure"

# 3. Refactor code file
cat script.py | gemini "Refactor with type annotations and docstrings" > script_clean.py

Troubleshooting & FAQ