Tags
12 posts
MCP has crossed 97 million monthly downloads and become the de facto standard, but there is no control layer governing which tools agents call and how often. The MCP Gateway pattern addresses this gap.
Solve the core challenges of deploying AI agents to production in 2026 with 9 battle-tested design principles from arXiv research, presented from an Engineering Manager perspective.
Google Research's 180-configuration experiment exposes the multi-agent paradox: 39–70% degradation on sequential tasks, 17.2× error amplification, and what it means for your architecture.
Analyzing the full prompt reprocessing issue when running Claude Code with local LLMs. Learn about KV cache invalidation mechanics and developer tool design lessons.
Analyzing a real implementation of MIT's RLM paper in coding agents. Learn how recursive self-invocation overcomes context limits and boosts single model performance by 91% from an engineering perspective.
When running multiple AI agents like Claude and Codex, task routing is the hardest challenge. It mirrors how engineering managers delegate work.
Deep Agents: 100-step tasks, explicit planning, auto-recovery for AI agent 2.0. Applies LangChain and Philipp Schmid principles to optimize agent architecture.
Apply MCP Code Execution patterns to reduce Claude Code token usage by 98.7% and boost speed by 60%. Step-by-step guide with .claude/ directory structure examples.
How a 17-agent blog automation system used metadata-first 3-tier design to cut tokens 60-70%, slash annual cost 71% ($5.72 to $1.65), and save 364 hours a year.
Removing the bloated recommendations.json and embedding recommendation data directly in frontmatter eliminates runtime file I/O by 100%. Here is our V3 system journey and results.
Eliminated 78K tokens and reduced execution time 99% using metadata caching and hash-based updates. Guide to building a near-zero cost recommendation system.
Orchestrate Architecture, Coding, Testing, Security, and DevOps agents to build production-grade applications