AI Engineering Podcast MCP as the API for AI‑Native Systems: Security, Orchestration, and Scale
28 snips
Dec 16, 2025 Craig McLuckie, co-creator of Kubernetes and CEO of StackLock, dives into the pivotal role of the Model Context Protocol (MCP) as the API layer for AI-native applications. He discusses the importance of securing AI agents through optimized MCP deployments and highlights common adoption pitfalls like tool pollution and security risks. Craig also stresses the need for continuous evaluations in stochastic systems and shares insights on ToolHive's innovative approach to orchestration and semantic search for better developer experiences.
AI Snips
Chapters
Transcript
Episode notes
MCP As The API For AI-Native Apps
- MCP functions as the API layer for AI-native applications, making the model the working view for modern apps.
- Treat MCP as the bridge that renders context in a model-consumable way and enables new multimodal experiences.
Avoid Running Unvetted MCP Packages
- Avoid NPX-installing random MCP packages with wide privileges because many are malicious and exploits are rising.
- Prefer hardened or curated MCP surfaces and SaaS integrations to reduce immediate risk.
Front Semantic Search For Tool Selection
- Put semantic search in front of tool sets to improve tool selection and reduce token costs.
- Return a top-K list so modest models can achieve high selection accuracy without using frontier models every time.