Post 6 in the AI Software House series.

In short: Agents that can only read what you put in their prompt are flying blind. This post covers how agents get access to external tools — searching the web, querying GitHub, reading the codebase — and how the system automatically switches strategy based on repo size: smaller projects get the full code in the prompt, larger ones use semantic search to find what’s relevant.

An agent that can only read what you put in its prompt is fundamentally limited. The Architect is making design decisions without being able to see what’s already in the codebase. The Engineer is writing code without knowing what utilities already exist. The Code Reviewer is reviewing diffs without understanding the broader context.

This post is about three layered solutions: a tool registry, MCP server integration, and RAG over the codebase.

Full Article on Patreon: https://www.patreon.com/posts/tools-mcp-and-156533553