Developer Argues MCPs Are an Expensive Mistake, Code-Based Tool Calling is the Future
generativist · x · 2026-08-06
After experimenting with custom agent harnesses, a developer expressed strong skepticism towards MCPs (Model Context Protocol), calling them and custom tools an expensive mistake. They argue that LLMs inherently want to write code, and forcing them into alternative tool-calling methods goes against their nature and sacrifices quality. The developer concludes that programmatic tool calling—where tool calls are executed as code—is the optimal way forward.
More from coding & agent
- AI Agents Caught Tampering With Memory Files, Security Researcher Admits — moyix · 2026-08-06
- SKILL-KD: Contrastive Skill Distillation for Weaker LLM Agents — ZhejiangUniversity · 2026-08-06
- OneDayAgent: A Harness for Long-Horizon Autonomous Agents — zjunlp · 2026-08-06
- Developer Seeks Growth Advice for MCP-Based AI Meeting Scheduler — Early_Ad8608 · 2026-08-06
- AI Agents Replicate ICML Papers: Exposing Hidden Compute Costs and Flaws — MengdiWang10 · 2026-08-06
- Open-Source AI Skill: Generate Multi-Style ATS Resumes via Interview-Style Q&A — vista8 · 2026-08-06