Articles from philschmid.de
8 kept
- MCP is Not the Problem, It's your Server: Best Practices for Building MCP Servers
MCP servers fail because they are built like REST APIs instead of as user interfaces for AI agents, requiring outcome-oriented design.
- Building Agents with the Gemini Interactions API
Building an agent with the Gemini Interactions API requires only an LLM loop, tool definitions, and server-side state management in under 100 lines of Python.
- Introducing MCP CLI: A way to call MCP Servers Efficiently
mcp-cli reduces MCP-related token usage by 99% through dynamic discovery, letting AI agents load only needed tool definitions instead of all upfront.
- The importance of Agent Harness in 2026
By 2026, agent harnesses will become essential infrastructure for building reliable AI systems that can execute complex, multi-day tasks.
- 8 Predictions for 2026. What comes next in AI?
The author predicts generative UI, edge agents, smart home context, agent harnesses, vibe coding as engineering, human-signed content, and biometric verification will define 2026.
- Context Engineering for AI Agents: Part 2
Manus and LangChain share production strategies for AI agents: compact context to prevent rot, isolate sub-agent contexts, keep tool sets small, and treat agents as tools rather than building complex scaffolding.
- Why (Senior) Engineers Struggle to Build AI Agents
Traditional software engineering's deterministic mindset clashes with the probabilistic nature of AI agents, requiring engineers to embrace ambiguity, text as state, and evals over tests.
- Gemini 3 Prompting: Best Practices for General Usage
This guide provides best practices for prompting Gemini 3, emphasizing clarity, structure, reasoning, and agentic tool use to maximize performance across domains.