gemini

Automate large-context codebase exploration with full event tracking via Gemini CLI.

3|3|Updated Dec 16, 2025
One-click install
npx skills add https://github.com/Shakes-tzd/htmlgraph --skill gemini-shakes-tzd
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: gemini
Source: https://github.com/Shakes-tzd/htmlgraph/tree/main/packages/claude-plugin/skills/gemini
Command: npx skills add https://github.com/Shakes-tzd/htmlgraph --skill gemini-shakes-tzd

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a structured way to orchestrate large-context research tasks by leveraging Google's Gemini CLI with HtmlGraph's full event tracking, enabling observability and provenance across complex explorations.

Core Features & Use Cases

  • Full Event Tracking: GeminiSpawner records parent/child events and subprocess activities for complete traceability during large-context research.
  • Flexible Execution Patterns: Choose between direct Gemini CLI invocation or Task(subagent_type="Explore") fallback, depending on availability and requirements.
  • Multimodal & Large Context: Supports code, documents, images, and PDFs with extensive context windows for deep analysis.
  • Real-World Use Case: A researcher analyzes a monorepo to identify architecture patterns and extract design decisions, saving results to HtmlGraph spikes.

Quick Start

Use the skill to start a large-context exploration with full event tracking. Example: Skill(skill=".claude-plugin:gemini", args="Analyze authentication patterns across the codebase") If Gemini CLI is not installed, the system automatically falls back to Task(subagent_type="Explore", prompt="Analyze authentication patterns in the codebase").

Frequently Asked Questions about gemini

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

FAQPage Schema
How do I track events during large-context codebase exploration?

Large-context codebase exploration tracks events by recording parent and child subprocess activities, ensuring complete traceability and reproducible provenance across complex research tasks.

Can I analyze code and documents in the same research task?

Multimodal research tasks support analyzing code, documents, images, and PDFs together within extensive context windows for deep architectural and documentation review.

What happens if the Gemini CLI is not installed for my research task?

If the Gemini CLI is not installed, the system automatically falls back to a Task subagent invocation to continue the large-context exploration without interruption.

How do I identify architecture patterns across a monorepo?

To identify architecture patterns across a monorepo, automate large-context exploration to extract design decisions and save the analysis results as tracked event histories.

Does this approach support reproducible provenance for code analysis?

Reproducible provenance for code analysis is supported by maintaining parent links and full event histories during multimodal research and architecture exploration.

What is the best way to save large-context research results?

The best way to save large-context research results is storing them as tracked spikes within an event-tracking system, preserving observable event histories for later review.