MCP ecosystem report: 54% of servers have zero community adoption
HM Cheng's June 2026 MCP Ecosystem Report, drawn from a live database of 39,762 active MCP servers, finds that 54% of tools score Grade C — solid code quality but no community signal, making them invisible to AI agents.
Score breakdown
The rebuilt scoring model replaces a system that compressed 85.7% of tools into a single grade, giving the ecosystem its first meaningful quality differentiation signal for identifying which MCP servers are actually discoverable by AI agents.
- 01Total MCP servers indexed: 39,762, all active within the last 30 days
- 0254.0% of servers are Grade C — solid code quality but zero community adoption
- 03Grade B+ (top tier) accounts for only 2.8% of servers (1,123 tools)
The June 2026 MCP Ecosystem Report, authored by HM Cheng and auto-generated by Agent Tool Intelligence, draws on a live database of 39,762 MCP servers, all of which were active (i.e., had a push within the last 30 days). The grade distribution reveals a heavily middle-weighted ecosystem: 54.0% of servers hold a Grade C ("average — good foundation, no community signal"), 18.1% Grade B, 13.8% Grade C+, 10.9% Grade D, 2.8% Grade B+, and just 0.4% Grade F. The central insight is that the majority of MCP tools have solid underlying code quality but no community adoption, which makes them invisible to AI agents.
The report also details a significant overhaul of the scoring methodology.
The report also details a significant overhaul of the scoring methodology. The previous static-analysis-only engine scored 85.7% of all tools as Grade B, providing no meaningful differentiation across the ecosystem. The replacement is a three-dimensional additive model that layers a Quality Score with a Community Bonus (incorporating signals such as GitHub stars and activity) and a Trust Bonus (incorporating official status and trust data). The report notes that the path from Grade C to Grade B is concrete: achieving 10 or more GitHub stars while remaining active is the described route.
Key facts
- 01Total MCP servers indexed: 39,762, all active within the last 30 days
- 0254.0% of servers are Grade C — solid code quality but zero community adoption
- 03Grade B+ (top tier) accounts for only 2.8% of servers (1,123 tools)
- 04Grade F (critical) accounts for 0.4% of servers (160 tools)
- 05The old scoring engine rated 85.7% of all tools as Grade B, making differentiation useless
- 06The new model is a three-dimensional additive system: Quality Score + Community Bonus + Trust Bonus
- 07The described path from Grade C to Grade B is earning 10+ GitHub stars while staying active
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Jun 12, 2026 · 10:05 UTC. How this works →