Unverified50% confidenceFactExact time
20 sources gave specific numbers for MCP tool counts, including Workato recommending 5-8, OpenAI recommending no more than 10 per namespace and no more than 20 in total, and Claude documentation recommending 30-50, but only 3 of them came from the authors' own measurements.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/14/2026
First Seen
Valid until: 1/12/2027
Sources
Related Entities
Related Claims
UnverifiedMultiple papers have specifically measured the quality of tool descriptions, and after analyzing approximately 100 MCP servers, researchers concluded that most descriptions are written very poorly.69% similarUnverifiedThis benchmark uses 10 original short documents of 200 to 250 words each, with 18 planted anomalies, 37 scoring entries in total (18 dirty and 19 clean), testing 13 models with 74 prompts per model.65% similarUnverifiedIn a test that used Traework to fully simulate a product development workflow, the tool retrieved more than 100 sources during the research phase, but after manual verification only about 40 were truly usable, while the other 60 needed to be discarded or re-investigated.63% similarUnverifiedA hand-crafted library of 22 calculators achieved perfect accuracy on the 440 cases it supported, but abstained on the remaining cases, yielding an overall coverage rate of only 40.0%.63% similarUnverifiedOpenAI's published blog posts only tested up to around 16 agents and did not push experiments to a scale of ten thousand agents.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1031408API
curl https://kongchang.com/api/v1/knowledge/claims/1031408MCP
get_claim(id=1031408)