MCP Trust Infrastructure

Know what your agent
connects to before it does.

Arclan is a machine-evaluated registry of MCP servers. Every server is validated, scored, and monitored continuously — so your agent has a trust signal before it makes a single tool call.

Query the registry How it works
—
Servers indexed
—
Active & validated
10s
Handshake timeout
24h
Revalidation cycle
Architecture

Three components.
One trust signal.

01
Discovery crawler
Continuously indexes GitHub, npm, PyPI, Smithery, and well-known endpoints. New servers enter a candidate queue the moment they appear.
02
Validation engine
Each server undergoes a full MCP handshake — protocol negotiation, tools/list enumeration, latency measurement — under a 10-second hard timeout.
03
Registry API
Query by capability, minimum score, or transport type. Every response includes a ready-to-paste MCP config snippet your agent can use immediately.
Live API

Query the registry
right now.

The Arclan registry exposes a simple REST API. Select an endpoint below and run it against the live registry.

GET /api/registry List servers
GET /api/stats Global counts
GET /api/registry/:id Server detail
POST /api/submit Submit a server
GET /api/registry 
# Query active servers with minimum score 70
$ curl -s https://arclan.ai/api/registry \
    -G \
    --data-urlencode "status=active" \
    --data-urlencode "min_score=70" \
    --data-urlencode "limit=5"
200 OK 

          

        

      

    

  







  

    
Trust model

    
Not a directory.
A trust layer.

    

      

        

          Existing MCP directories are static lists — human-curated, not machine-evaluated. Arclan validates every server mechanically and scores it on a rolling basis. The score decays. Status transitions. Dead servers are marked dead.
        

        

          

            
→

            

              
Composite score

              
70% weight on the most recent validation result, 30% on the 30-day rolling history. Score range 0–100.

            

          

          

            
→

            

              
State machine

              
Servers move through CANDIDATE → ACTIVE → DEGRADED → DEAD → ZOMBIE states based on observed behaviour, not manual review.

            

          

          

            
→

            

              
Agent telemetry

              
Wild agents report operational outcomes back to the registry via POST /api/report — crowdsourced trust signals that compound over time.

            

          

        

      


      
      

        
Live score samples


        

          beagel.io/mcp
          90
        

        

          

        


        

          example-mcp.dev
          61
        

        

          

        


        

          stale-tools.io/mcp
          12
        

        

          

        


        

          

            
Transport

            
SSE

          

          

            
Latency

            
308ms

          

          

            
Status

            
ACTIVE