Napačna izbira? Nič za to! Ponujamo možnost vračila v 30 dneh
Z darilnim bonom ne morete zgrešiti. Obdarovanec lahko v zameno za darilni bon izbere karkoli iz naše ponudbe.
30 dni za vračilo blaga
Your AI Agent is Powerful. But It Can't See Your Private Data.
The Gemini CLI marks a pivotal shift from passive chat to active execution. But its built-in tools, while capable, are generic. They can't access your company's private Jira instance, query your internal databases, or automate pull requests in your GitHub Enterprise repository . This is the wall that separates a helpful assistant from a true, mission-critical collaborator.
This book is not about prompting. It is about building.
This is the definitive developer's guide to breaking that wall. "Extending Gemini CLI" is your practical, hands-on playbook for mastering the Model Context Protocol (MCP)-the "API for AI" that securely connects the Gemini agent to your private, proprietary, and mission-critical systems .
Stop wishing your AI could "just check the internal wiki". This book shows you how to build the tools that make it possible.
Inside, you will master:
The MCP Architecture: Go from user to builder. Understand the complete MCP architecture, from tool discovery and communication to designing schemas with the OpenAPI specification .
Core Agentic Design: Learn the hard-won principles for building reliable tools. Master concepts like "Tools, Not Chatbots" (structured JSON vs. text), Atomicity (small, composable tools), and writing effective descriptions-the real prompt engineering .
Production-Grade Authentication: Secure your tools from day one. Implement the two essential patterns: static API keys for internal services and user-specific OAuth 2.0 for production-grade, delegated access.
Expert Case Studies: Build three high-value, enterprise-ready tools from scratch:
GitHub Enterprise Tool: Automate pull requests and read issue data .
Jira Tool: Summarize sprints and report on bug status .
Private Database Tool: Learn to avoid the dangerous "Text-to-SQL" Anti-Pattern and build safe, parameterized query tools for your internal data.
The Deployment Blueprint: Containerize your MCP server with Docker and deploy it to AWS Lambda, Google Cloud Run, or a private VPC .
Advanced Agentic Architecture: Master tool chaining, observability, and designing for failure to build a resilient, multi-agent ecosystem .
This book is for the engineer, architect, or developer who sees AI not as a black box, but as a collaborator to be empowered. If you're ready to build the next generation of truly integrated AI agents, this is your starting point.
Let's build the agentic future.