An analysis of 119,510 signed MCP server results from August to October 2026 reveals insights into server uptime and tool-list drift. The data, collected by probing public remote MCP servers, indicates that approximately 60.3% of observed endpoints had signed, full coverage results. The study defines uptime based on server responsiveness, noting that changes in how errors are categorized (e.g., HTTP 5xx responses) may slightly skew earlier data towards appearing more available. AI
IMPACT Provides insights into the operational stability and configuration consistency of MCP servers.
RANK_REASON Analysis of a large dataset of server results to identify trends. [lever_c_demoted from research: ic=1 ai=0.1]
- 23:00 UTC
- August 20, 2026
- August 22, 2026
- MCP
- October 20
- October 2026
- October 21, 2026
- October 22, 2026
- September 23
- September 27
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →