Stop Giving Your Agent Every Tool
Large tool catalogs break agent context. Tool search fixes that by letting agents discover and load only what they need.
I walk through the tool-context problem I hit in Emma, the Anthropic tool-search pattern for large tool surfaces, and why I replaced my custom version with Mastra’s built-in ToolSearchProcessor. You’ll see the full search, load, and call loop in Mastra Studio, from an agent with no direct tools to a targeted tool invocation.
Building an AI agent?
I help teams design and ship agentic systems — from architecture to production.
See how I can help
I Added ACP to My Mastra Agent So It Can Work in Repos
ACP is the layer that lets my Mastra agent hand real repo work to Claude Code instead of stopping at advice.
Stop Letting AI Agents Run the Whole Workflow
One inbox agent should not classify, research, score, route, and draft replies in one loose loop.
Harness Engineering: 4 Levers to Diagnose Any AI Agent
Most agent failures aren't model failures. They're harness failures. Here's the 4-lever framework I use to diagnose what broke.
Building Approval Gates AI Agents Can't Route Around
How to wire human-in-the-loop on tool calls — and why system prompt instructions like "always ask before sending" don't actually hold.
Get new videos and posts by email
Weekly videos on AI engineering, plus deeper dives in the newsletter.
Occasional emails, no fluff.