Google Gemini File Search as a Gemini CLI Extension

Description

In this demo, we take Gemini File Search — Google’s cloud-native Retrieval Augmented Generation (RAG) service — and incorporate it into Gemini-CLI as an extension for real-time hallucination free AI. You’ll see how to: ✅ Upload and index PDFs, JSON, and CSV files into Gemini File Search ✅ Fix MIME type errors (application/json → text/plain) for full compatibility ✅ Ask grounded, context-aware questions powered by Gemini 2.5 Flash ✅ Combine Ansible, pyATS, and Cisco data with Gemini’s cloud RAG pipeline ✅ Turn local automation files into searchable, explainable AI knowledge This setup abstracts RAG completely — no vector DB, no embeddings, no retrievers — just clean cloud-hosted File Search directly accessible via Gemini MCP tools. 📂 Tools shown: /file_search:upload_and_index /file_search:import_file /file_search:query_file_search /file_search:list, /get, /delete 🌐 Example Use Cases: Understand Jinja2 templates using AutomateYourNetwork.pdf Analyze show ip interface brief outputs in JSON/CSV Build end-to-end network automation knowledge bases with Gemini 🔧 Tech Stack: Python + FastMCP + Google GenAI SDK + Gemini 2.5 Flash Runs locally or in the cloud as a full MCP extension for the Gemini CLI. 🎥 Subscribe for more AI + Network Automation deep dives: Gemini CLI | MCP | File Search | pyATS | Ansible | Selector AI The GitHub repo: https://github.com/automateyournetwork/GeminiCLI_File_Search_Extension Install it: gemini extensions install https://github.com/automateyournetwork/GeminiCLI_File_Search_Extension.git