I tested 1,682 public MCP servers: 51% have tool inputs with no description
abhiramreddykalluru · reddit · 2026-11-14
The author runs a free MCP server tester that runs every server in the official registry through the same safe test. 1,682 produced a score (another 1,459 require login and couldn't be scored externally). The most common issue is boring: 51% of servers have tool inputs with no description — the AI must guess what limit means and often guesses wrong.
Remaining issues by prevalence: 26% handle bad input poorly (crash, hang, or unrecoverable errors); 20% have tools that fail even with good input; 19% don't mark read-only vs. mutating tools; 10% are slow enough that some clients give up. Only 150 of 1,682 scored a perfect 100.
The silver lining: the top problem is the easiest fix. One founder added one-sentence descriptions to 14 tools and turned fixed choices into enums, going from 93 to 100 the same day. The author's advice: open your tool list and give every input one plain sentence — what it is, what values it takes, the default.
More from coding & agent
- Money-saving trick: let a cheap model do the work, then ask it to summon a pricier model to review — vista8 · 2026-11-14
- Open-source personal agent The Machine drafts everything but never sends without your OK — markloperman · 2026-11-14
- Open-source Orinyx tests healthcare AI agents against FDA labels, no LLM in the verdict — Practical_Green5775 · 2026-11-14
- Solo AI system sold for $18,880 watches 1,300 execs and surfaces $50k–$200k deals — aryanXmahajan · 2026-11-14
- AAA game dev builds MAC, a multi-agent council where GPT, Claude and Codex review each other — Hirox-rd · 2026-11-14
- Perplexity team teases minimalist LLM trainer built with coding agents, open source soon — BanghuaZ · 2026-11-14