Adoption
Which tools carry the product, and which ones are never selected?
MCP tool usage analytics
Know which MCP tools agents actually use
Measure adoption, selection, errors, latency, retries, and completion. Improve the tools that make a workflow work—not the ones that only look good in a schema.
tool adoption · last 7 dayslive
search_docs14,208 calls · 100%
create_issue9,841 calls · 69%
run_query6,502 calls · 46%
deploy_service0 calls · unused
8
tools
3
need work
75%
completion
A tool is more than its schema
The usage signal tells you what to improve.
A tool can be valid, documented, and deployed—and still be ignored by agents. Connect selection to the result of the workflow.
Selection
Do agents understand the tool names, descriptions, and input schemas?
Errors
Which tools fail, retry, or return application errors inside a 200 response?
Completion
Do the calls add up to a completed workflow or an abandoned session?
From catalog to outcome
Find the leak between discovery and completion.
See where agents drop off after discovering your tools, which calls get retried, and whether a schema or dependency is costing you completed work.
Read MCP observabilityWorkflow drop-offlast 7 days
Connected100%
Discovered tools93%
Called a tool81%
Completed75%
75 of 100 sessions reached a result
MCP tool analytics
Build for the tools people use.
Start measuring your catalog with one server, one key, and one real workflow.