Engineering Blog

Insights from the lab

Deep dives on AI agents, Claude Code, MCP servers, automation, testing, and software architecture.

Showing 4 of 6 posts

May 15, 2025

Building Claude Code Extensions: Skills, Plugins, and Internal Marketplaces

Claude Code is extensible at every layer — custom slash commands, MCP servers, hooks, and full plugin systems. Here's how to build extensions your whole team can discover and use.

Testing AI Agents: A Practical Strategy for Reliable Agentic Systems

Unit tests don't cut it for agents that reason dynamically and call external tools. Here's the testing strategy we use to ship agentic systems with confidence — from evals to integration harnesses.

Agentic Workflow Patterns: Orchestrating AI Agents at Scale

Single-agent loops break down in complex systems. Here are the orchestration patterns — chains, fan-out, human-in-the-loop, and more — that make multi-agent pipelines reliable in production.

Arjun KayalMoni

Read article

Building an MCP Server in TypeScript: A Complete Guide

May 5, 2025

MCP Servers6 min read

Building an MCP Server in TypeScript: A Complete Guide

The Model Context Protocol (MCP) lets you give Claude and other LLMs access to custom tools, data sources, and APIs. Here's how to build, test, and deploy your own MCP server from scratch.

Arjun KayalMoni

Read article

Loading more…