Skip to main content
Insights shows what happens to your project in production: every run of a release through the REST API, MCP, or an n8n workflow. Runs you start in the editor and test runs are not included.

​
Open Insights

In the Blueprint Editor, click Insights in the view switcher of the top toolbar (Design | Playground | Insights). To return to editing, click Design. The rail on the left lists the sections: Insights is available to members of the workspace. People who open the project through a public link don’t see it.

​
Time window and version

At the top of each Analytics section you choose what to look at:
  • Time window — 24h, 7d (the default), or 30d
  • Filter by version — All versions, or one release. Use it to compare how a new release behaves against an earlier one.

​
Summary

The Summary shows the health of your project in the selected time window: The Insights Summary for the Loan Application Scorecard project — 45 runs, a 95.6% success rate, average duration, worst p95, the Outcomes bar, and the Entry points table with both public Blueprints
  • Runs — how many production runs there were
  • Success rate — the share of runs that succeeded, with the number of failed runs
  • Avg duration — the average run time
  • p95 duration — 95% of runs finished within this time. If the project has several public Blueprints, the tile is called Worst p95 and names the entry point with the slowest p95.
The Outcomes bar splits all runs into successes, errors, and timeouts. The Entry points table breaks the numbers down by public Blueprint: Runs, Success, Avg, p95, and Share of all runs.

​
Prod Runs

Prod Runs lists the 50 most recent production runs. Each run shows its status, its duration, and the release it ran on. Narrow down the list with the filters at the top:
  • Status — All, Success, Error, or Timeout
  • Search by correlation ID — find one specific run. Callers set the correlation ID with the X-Correlation-Id header, see Tag a run with a correlation ID.
  • The time window and the version filter described above
Click a run to expand it and see its correlation ID, its Input, and its Output — or the error it ended with. A production run of the Loan Application Scorecard found by its correlation ID APP-2026-1003 and expanded, showing the input, the output, and the Replay button

​
Replay a run

Click Replay on a run to open its recorded trace on the diagram of the release that ran. You can follow the exact path the run took, step by step. Replay does not run the Blueprint again; it shows what happened at the time. Click Runs above the diagram to return to the list. To share a replay with a colleague, click the link icon next to Replay (Copy link to replay).

​
Create a test case from a run

When a production run shows a case worth keeping, turn it into a test:
  1. Expand the run and click the clipboard icon next to Replay (Create test case from run).
  2. The dialog is prefilled with a name, the run’s inputs, and its result as the expected outputs. It names the test suite the case is added to — if the Blueprint has no test suite yet, Leapter creates one.
  3. Adjust the values if needed. Each expected output is either Checked — compared with the actual result when the test runs — or Ignored. Click the pill next to an output to switch.
  4. Click Add test case.
The Create test case from run dialog, prefilled from a production run and adding to the Lending decisions test suite, with the four inputs and four checked expected outputs Leapter opens the new test case in the Tests panel. You can capture runs that succeeded and runs that ended with an error your Blueprint raises on purpose. Runs that timed out or failed for technical reasons can’t be captured.

​
Logic Usage

Logic Usage paints production usage onto your logic. Choose a Blueprint with Select blueprint: Leapter shows its diagram as it was in the selected release and marks every node with the share of runs that reached it. A node shown in red failed in at least one run. Logic Usage for the Loan Application Scorecard on release v1.0 — every node carries the share of runs that reached it, for example 31% for a strong score, 53% for a marginal score, and 9% for a weak score Use it to find logic that production never reaches, or to see which branches carry most of the traffic. Logic Usage always shows one release at a time, because different releases can contain different logic. Pick the release with the version filter. When there are many runs in the time window, the percentages are calculated from the newest runs. Hover over the info icon next to the scale for details.

​
Releases

Releases lists every release of the project. The Latest release is the one production runs; older ones are listed under Earlier releases. Each entry shows the version name and description, its Content ID, who created it and when, the project version it was created from, and the number of Blueprints. Click Download to save the release package. See Releases to learn how to create one.

​
No data yet?

​
What to do next

  • Releases — create the release that production runs
  • Test Suites — run the test cases you captured from production
  • Web API (OpenAPI) — call your project from your applications