forayconsulting

Gemini CLI Integration Patterns

Development community intermediate

SOURCE https://github.com/forayconsulting/gemini_cli_skill

Description

Advanced patterns for orchestrating Gemini CLI effectively from Claude Code.

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

Repository README

This is the README for forayconsulting/gemini_cli_skill, shared by 3 entries in this directory. It describes the repository, not this entry specifically.

Gemini CLI Integration Patterns

Advanced patterns for orchestrating Gemini CLI effectively from Claude Code.

Pattern 1: Generate-Review-Fix Cycle

The most reliable pattern for quality code generation.

# Step 1: Generate code
gemini "Create [code description]" --yolo -o text

# Step 2: Have Gemini review its own work
gemini "Review [generated file] for bugs and security issues" -o text

# Step 3: Fix identified issues
gemini "Fix these issues in [file]: [list from review]. Apply now." --yolo -o text

Why It Works

Different "mindset" for generation vs review
Self-correction catches common mistakes
Security vulnerabilities often caught in review phase

Example

# Generate
gemini "Create a user authentication module with bcrypt and JWT" --yolo -o text

# Review
gemini "Review auth.js for security vulnerabilities" -o text
# Output: "Found XSS risk, missing input validation, weak JWT secret"

# Fix
gemini "Fix in auth.js: XSS risk, add input validation, use env var for JWT secret. Apply now." --yolo -o text

Pattern 2: JSON Output for Programmatic Processing

Use JSON output when you need to process results programmatically.

gemini "[prompt]" -o json 2>&1

Parsing the Response

// In Node.js or with jq
const result = JSON.parse(output);
const content = result.response;
const tokenUsage = result.stats.models["gemini-2.5-flash"].tokens.total;
const toolCalls = result.stats.tools.byName;

Use Cases

Extracting specific data from responses
Monitoring token usage
Tracking tool call success/failure
Building automation pipelines

Pattern 3: Background Execution

For long-running tasks, execute in background and continue working.

# Start in background
gemini "[long task]" --yolo -o text 2>&1 &

# Get process ID for later
echo $!

# Monitor output incrementally with BashOutput tool

When to Use

Code generation for large projects
Documentation generation
Running multiple Gemini tasks in parallel

Parallel Execution

# Run multiple tasks simultaneously
gemini "Create frontend" --yolo -o text 2>&1 &
gemini "Create backend" --yolo -o text 2>&1 &
gemini "Create tests" --yolo -o text 2>&1 &

Pattern 4: Model Selection Strategy

Choose the right model for the task.

Decision Tree

Is the task complex (architecture, multi-file, deep analysis)?
├── Yes → Use default (Gemini 3 Pro)
└── No → Is speed critical?
    ├── Yes → Use gemini-2.5-flash
    └── No → Is it trivial (formatting, simple query)?
        ├── Yes → Use gemini-2.5-flash-lite
        └── No → Use gemini-2.5-flash

Examples

# Complex: Architecture analysis
gemini "Analyze codebase architecture" -o text

# Quick: Simple formatting
gemini "Format this JSON" -m gemini-2.5-flash -o text

# Trivial: One-liner
gemini "What is 2+2?" -m gemini-2.5-flash -o text

Pattern 5: Rate Limit Handling

Strategies for working within rate limits.

Approac