gemini

What this does

Provides access to Google Gemini AI models for code analysis, brainstorming, file analysis, and general questions. Supports multiple brainstorming methodologies and structured change mode for code edits.

Setup

mcpc connect <your mcpc config>:gemini-cli @gemini

Tools

ask-gemini

Ask Gemini a question or request analysis. Supports file references via @ syntax.

Parameters:

Parameter	Type	Required	Description
prompt	string	yes	Analysis request. Use @filename to include files.
model	string	no	Model to use (default: the CLI's current flagship model; check mcpc @gemini docs for valid snapshot names).
sandbox	boolean	no	Run in isolated sandbox for risky operations (default: false)
changeMode	boolean	no	Return structured edit suggestions (default: false)

Examples:

mcpc @gemini tools-call ask-gemini prompt:="@main.go explain the architecture and suggest improvements"
mcpc @gemini tools-call ask-gemini prompt:="Compare REST vs GraphQL for a blog API" model:="<flash-tier model name>"
mcpc @gemini tools-call ask-gemini prompt:="@config.yaml refactor to use environment variables" changeMode:=true

brainstorm

Generate ideas using creative frameworks with domain context and analysis.

Parameters:

Parameter	Type	Required	Description
prompt	string	yes	Challenge or question to explore
methodology	enum	no	divergent, convergent, scamper, design-thinking, lateral, auto (default)
domain	string	no	Context: software, business, creative, research, product, marketing
constraints	string	no	Limitations (budget, time, technical, legal)
existingContext	string	no	Background info or previous attempts
ideaCount	integer	no	Target number of ideas (default: 12)
includeAnalysis	boolean	no	Include feasibility/impact analysis (default: true)
model	string	no	Model override

Examples:

mcpc @gemini tools-call brainstorm prompt:="API design for multiplayer game" domain:="software" methodology:="design-thinking"
mcpc @gemini tools-call brainstorm prompt:="Reduce churn in SaaS" domain:="business" methodology:="scamper" constraints:="low budget"

fetch-chunk

Retrieve cached chunks from a changeMode response.

Parameters:

Parameter	Type	Required	Description
cacheKey	string	yes	Cache key from initial changeMode response
chunkIndex	number	yes	1-based chunk index

ping

Echo test. Parameters: prompt (optional string).

Help

Show help information. No parameters.

When to use

• Need a second opinion from a different AI model
• Brainstorming with structured creative frameworks (SCAMPER, Design Thinking, etc.)
• Analyzing files with @filename syntax
• Generating structured code change suggestions via changeMode
• Quick questions where a different model perspective helps

Workflow

Code review with file context

mcpc @gemini tools-call ask-gemini prompt:="@file.go review for bugs, performance issues, and suggest improvements"

Structured code changes

1. mcpc @gemini tools-call ask-gemini prompt:="@config.yaml add support for environment variables" changeMode:=true
   # → Returns structured edit with cacheKey if response is chunked
2. mcpc @gemini tools-call fetch-chunk cacheKey:=<key> chunkIndex:=2 → get next chunk if needed

Brainstorming session

1. mcpc @gemini tools-call brainstorm prompt:="..." domain:="software" methodology:="scamper" ideaCount:=8
2. Review generated ideas with feasibility analysis
3. mcpc @gemini tools-call brainstorm prompt:="Refine top 3 ideas" methodology:="convergent" existingContext:=<previous results>

Best practices

• Use @filename syntax to give Gemini file context
• Use the faster/cheaper flash-tier model for simple queries; the pro-tier flagship for complex analysis (model snapshot names change - query the CLI, don't hardcode)
• Enable sandbox:=true when testing potentially risky code
• For brainstorming, let methodology:="auto" unless you need a specific framework
• Use changeMode:=true when you want structured, applicable edit suggestions