#mcp

2 articles

Architecting Zero-Side-Effect Evals for AI Agents and MCP Tooling

Learn how to build deterministic, zero-cost evaluation harnesses for AI agents and Model Context Protocol (MCP) servers without mutating production data.

Thanga·

8 min read

ai-agentstesting

Architectural Deep-Dive: Building Hermetic Test Harnesses for AI Agent Skills and MCP Servers

Learn how to build transparent mock layers and deterministic AX evaluation suites for tool-calling AI agents and MCP servers without mutating production data.

Thanga·

7 min read