MCPs are becoming the API layer for AI agents.
But in production, they're a black box : silent failures, error mismatches, and latency spikes.
Say hello to Scope - Observability for MCP servers.
๐ฆ Install with a single command
๐ Initialize with one line of code
๐ Track requests, latencies, and tool calls in real time
We're in public beta. Start tracking your MCP today: ๐ https://t.co/SHoYKKRUL2
MCPs are becoming the API layer for AI agents.
But in production, they're a black box : silent failures, error mismatches, and latency spikes.
Say hello to Scope - Observability for MCP servers.
๐ฆ Install with a single command
๐ Initialize with one line of code
๐ Track requests, latencies, and tool calls in real time
We're in public beta. Start tracking your MCP today: ๐ https://t.co/SHoYKKRUL2
@Fispiy@_adishj@shivvtrivedi dude lol.
cracked team man!
I am building a SDK to monitor your MCP.
Traffic/errors/latencies/etc.
Let me know if @shivvtrivedi needs a break.
@sonnet_4_5_ Tracking calls across multiple agents is on the roadmap.
We are starting with tracking just MCP traffic/errors/latencies.
Curious to know about your use-case!
@yousefhll this is pure madness custom UI per user is madness.
But shouldn't the user be handed what to do.
If only, you could somehow detect which parts the user uses most and frame the UI accordingly.
But, beautiful tool indeed!