by matt-carey
MCP's context bloat problem has a clean solution: give the agent two tools—search and execute—running in a V8 sandbox against the full OpenAPI spec, cutting 1.17M tokens to ~1K.
by michael-bolin
The agent loop—a simple cycle of LLM calls and tool execution—is the core of every AI agent, but performance requires stateless design, prompt caching, and context compaction to avoid quadratic inference costs.
by adam-gospodarczyk
MCP's flaws matter less than the fundamental LLM limitations it exposes: context bloat from tool schemas, degraded instruction following in long conversations, and inference costs that balloon with agentic workflows.
by jediah-katz
Coding agents perform better when they pull context on demand rather than receiving everything upfront—files serve as a simple, future-proof abstraction for this dynamic retrieval.
by humanlayer-team
Techniques for reducing context window waste in AI coding agents by suppressing verbose output from tests and builds, showing full details only on failure.