collaborating-with-gemini

Delegate coding tasks to the Gemini CLI and return structured JSON.

4|Updated Feb 21, 2026
One-click install
npx skills add https://github.com/Fanceir/paper-skill-latex --skill collaborating-with-gemini-fanceir
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: collaborating-with-gemini
Source: https://github.com/Fanceir/paper-skill-latex/tree/main/.codex/skills/collaborating-with-gemini
Command: npx skills add https://github.com/Fanceir/paper-skill-latex --skill collaborating-with-gemini-fanceir

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

Provide a reliable, low-token bridge to the Gemini CLI so engineers can delegate prototyping, debugging, and code review tasks while preserving session continuity and structured outputs.

Core Features & Use Cases

  • Session-aware collaboration: Start and resume multi-turn Gemini sessions with SESSION_ID to keep context across follow-ups.
  • Safe, structured outputs: Returns JSON containing success flags, SESSION_ID, agent messages, and optional full message traces, and enforces unified-diff output for code changes.
  • Workspace-centric handoffs: Optimized for file/line references and repository-root execution to avoid pasting large code blocks and to support reproducible local verification.

Quick Start

Run the gemini bridge with a concise instruction, set the workspace root, and request a unified diff patch to receive a JSON response containing SESSION_ID and agent messages.

Frequently Asked Questions about collaborating-with-gemini

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I delegate coding tasks to the Gemini CLI for code review and debugging?▼

You can delegate coding tasks to the Gemini CLI by running the bridge script with a concise instruction and specifying your workspace root. It returns structured JSON containing SESSION_ID and agent messages for prototyping and code review.

Can I resume a previous Gemini CLI session for multi-turn debugging?▼

Yes, you can resume multi-turn Gemini CLI sessions using the SESSION_ID. This session-aware collaboration preserves context across follow-ups, allowing continuous debugging and prototyping without losing previous conversation history.

How does the Gemini CLI handle code changes and patch outputs?▼

The Gemini CLI enforces unified-diff output for code changes. This workspace-centric approach uses file and line references to avoid pasting large code blocks, ensuring reproducible local verification directly within your repository root.

Do I need to paste large code blocks when using the Gemini CLI for code review?▼

No, you do not need to paste large code blocks. The workspace-centric handoff uses file and line references from a specified workspace root, allowing the Gemini CLI to review code directly from your repository.

What format does the Gemini CLI return for structured outputs and error reporting?▼

The Gemini CLI returns structured JSON containing success flags, SESSION_ID, agent messages, and optional full message traces. This structured output ensures clear error reporting and reliable parsing of agent responses.

Are sandboxed runs supported when executing Gemini CLI tasks from a workspace root?▼

Yes, optional sandboxed runs are supported when you execute Gemini CLI tasks. This satisfies the requirement to run the Gemini CLI from a specified workspace root safely while performing prototyping and debugging.