Stats
Actions
Tags
Help us improve
Share bugs, ideas, or general feedback.
From AI-skill
Runs Google's Gemini CLI for code generation, review, analysis, web research via Google Search, and codebase architecture investigation. Use for second AI opinions, real-time web data, or parallel code tasks.
npx claudepluginhub caidish/cai-tools --plugin AI-skillHow this skill is triggered — by the user, by Claude, or both
Slash command
/AI-skill:gemini-cliThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
| Use Case | Why Gemini |
Delegates tasks to Gemini CLI for large-context analysis like broad codebase reviews or long-document processing. Activates on explicit requests such as 'use gemini' or 'delegate to gemini'.
Invokes Google Gemini CLI for complex reasoning, research, and AI tasks in headless mode. Supports preview models, fallbacks, and session continuation.
Invokes Google Gemini CLI for second opinions, architectural advice, code reviews, security audits, and debugging using 1M+ context window in Claude Code workflows.
Share bugs, ideas, or general feedback.
| Use Case | Why Gemini |
|---|---|
| Current web information | google_web_search - real-time Google Search |
| Codebase architecture analysis | codebase_investigator - deep analysis tool |
| Second opinion / code review | Different AI perspective catches different bugs |
| Parallel code generation | Offload tasks while continuing other work |
When NOT to use: Simple quick tasks (overhead not worth it), interactive refinement, context already understood.
command -v gemini--yolo) unless edits are necessary.AskUserQuestion before using --yolo or -s flags. These modes allow file writes or sandboxed execution - get explicit user approval first.-m, --model <MODEL> - Model selection-y, --yolo - Auto-approve all tool calls (enables writes)-s, --sandbox - Run in Docker isolation-o, --output-format <text|json> - Output formatgemini "prompt" -o json 2>/dev/null | jq -r '.response' to suppress stderr noise and extract the json response, unless specified by the user.YOLO mode does NOT prevent planning prompts. Use forceful language: "Apply now", "Start immediately", "Do this without asking for confirmation".
| Use case | Mode | Command pattern |
|---|---|---|
| Read-only analysis | read-only | gemini "..." -o json 2>/dev/null | jq -r '.response' |
| Apply local edits | write | gemini "..." --yolo -o json 2>/dev/null | jq -r '.response' |
| Sandboxed write | sandbox | gemini "..." --yolo --sandbox -o json 2>/dev/null | jq -r '.response' |
# Read-only
gemini "Review src/ for bugs" -o json 2>/dev/null | jq -r '.response'
# Write mode
gemini "Fix bug in file.py. Apply now." --yolo -o json 2>/dev/null | jq -r '.response'
# If redirection fails, wrap in bash -lc
bash -lc 'gemini "prompt" -o json 2>/dev/null | jq -r ".response"'
echo "follow-up" | gemini -r latest -o json 2>/dev/null | jq -r '.response'gemini --list-sessions-m gemini-2.5-flash for lower priority tasks.gemini --version, use --debug for details.