Description
Check whether the observed reasons match the jobs the tool is meant to support.
MCP tool analytics Know which tools people use. And what they use them for.
MCP tool analytics measures tool usage, adoption and reliability. Flowlines adds the reasons behind calls, so you can distinguish a popular capability from a tool that agents keep retrying.
100 analyzed MCP sessions per month, free. No credit card. Connect your assistant, then instrument your server.
One tool. Several very different jobs.
Select a use case to inspect an example of the wording behind the calls.
search_customers
Last 7 daysObserved calls360
Calls with a reason360
Grouped use cases3
Failed calls12
What people use it for
The tool stays the same. The job changes.
Example call reason
“Find the account so I can review its activity before the renewal meeting.”
Account lookup is the first step, not the finished brief.
Supplied contextNot hidden model reasoning
What is MCP tool analytics?
MCP tool analytics measures how tools exposed by an MCP server are used: call volume, participating users and sessions, failures, timing and the jobs behind those calls. It helps you decide which tools to improve, investigate or promote.
Flowlines' tool view includes a What people use it for analysis, with grouped call reasons and examples of their wording. That makes search_customers more informative than a line on a volume chart: the same operation may support renewal preparation, account lookup or duplicate-record investigation.
Put tool volume in context.
Read tool activity alongside the population and workflows it serves. A technically successful call is a reliability observation, not proof that the user finished their task.
| Measure | Use it to | Avoid concluding |
|---|---|---|
| Calls by tool | See where activity concentrates. | More calls always mean more value. |
| Users and sessions | Separate broad adoption from repeated use by a few people. | Every call represents a different customer. |
| Failed calls | Find operations and use cases needing investigation. | Every error causes the whole workflow to fail. |
| Call latency | Spot slow operations in the relevant path. | A faster call guarantees a better outcome. |
| Call reasons and examples | Understand the work the tool supports. | A tool name alone reveals the user's goal. |
| Repeated calls | Investigate loops or inefficient paths in a session. | Every repetition is a retry; pagination can be intentional. |
How do you find MCP tools nobody uses?
Compare the tools declared by your server with the tools called during a defined period. A declared tool with zero observed calls is unused in that window, provided your catalog and telemetry coverage are complete.
Flowlines shows observed tool activity and available contract information. If the integration only reports tools when they run, that call history cannot reveal tools that have never run. Verify the catalog before describing a missing tool as unused.
Zero calls does not tell you why. The tool may be seasonal, inaccessible to a client, poorly described or simply unnecessary. Similarly, a declared tool is not proof that a particular client discovered it. Validate those hypotheses before changing the product.
The busiest tool is not always the most useful.
Imagine search_customers is the busiest tool on a server. Some calls retrieve an account immediately. Others repeat because the result does not distinguish two companies with similar names. The total treats both patterns alike.
Group the observed reasons, inspect the corresponding sessions and check the returned result where captured. You can then decide whether the tool needs a clearer description, a more specific argument or a different response. Call frequency helps you find the question; context helps you answer it.
Did the tool change help?
Compare usage before and after a description, schema or release change with the same time windows and comparable traffic. Flowlines exposes contract changes and time-based analytics that support this investigation.
Look at calls and failures, then inspect affected sessions. A lower error rate with fewer users or less outcome coverage is not automatically an improvement. Treat the comparison as an investigation, not a controlled experiment.
Schema
Look for changed arguments, failed calls and repeated attempts around the new version.
Workflow
Compare the path people take and the available outcome evidence, not just the number of calls.
Connect your assistant. Instrument your MCP.
The Flowlines plugin gives your assistant the setup workflow. Connecting the assistant does not automatically instrument your MCP server.
- Connect FlowlinesUse the button to open the supported setup path in your assistant.
- Instrument the serverUse the plugin's MCP observability workflow. Review what will be captured before sending production data.
- Verify the first callsCheck tools, identity and session coverage. Add outcome context where supported and keep missing data visible.
MCP OpenTelemetry instrumentation guide ↗MCP observability product overview ↗The MCP analytics measurement guide ↗
MCP tool analytics: common questions
How do I track MCP tool usage?
Instrument calls with compatible telemetry, choose a time window and group by tool. In Flowlines, inspect tool activity alongside available users, sessions and call reasons.
How do I calculate MCP tool success rate?
Use calls with a known successful status divided by calls with a known result, and report unknown status separately. Label this call success, not task completion. State the time window and exclusions.
Can Flowlines show why agents choose a tool?
Flowlines can group the call reasons supplied by the integration and show examples. These are reported reasons, not access to a model's hidden reasoning, and missing context stays missing.
Can I compare tool usage before and after a release?
Yes. Use available contract history and comparable analytics windows, then inspect representative sessions. Differences in users, clients or telemetry coverage can affect the comparison.
Start with the traffic you already have
See the work behind the calls.
Connect Flowlines and take the next step with your assistant.