What are insights?
Insights are read-only, workspace-scoped analytics derived from your own submissions, issues, and processing jobs. They answer questions like how many assets did we process this month, which rules trip most often, what share of submissions pass clean, and how much review time are we saving. The workspace is implicit — every insights request is scoped to the workspace behind your API key, so there is no workspace parameter to pass.Sandbox submissions are excluded from every insights metric. Sandbox keys are
for integration testing only, so their canned results never count toward your
workspace’s counts, rates, or durations.
Endpoints
Each metric is its own endpoint, so an endpoint’s query parameters apply to exactly the stat it returns (no parameter that quietly does nothing).| Endpoint | Returns | Params |
|---|---|---|
GET /insights/assets/count | assets created per day | days, sidekickId |
GET /insights/assets/by-type | asset distribution by file type | sidekickId |
GET /insights/assets/violation-rate | share of assets with issues, per day | days, sidekickId |
GET /insights/issues/by-severity | issue counts by severity, per day | days, sidekickId |
GET /insights/issues/by-sidekick | issues grouped by sidekick | days |
GET /insights/issues/by-file-type | issues grouped by file type | days, sidekickId |
GET /insights/submissions/pass-rate | clean-completion rate, per day | days, sidekickId |
GET /insights/submissions/by-file-type-sidekick | success/failure by type × sidekick | days |
GET /insights/rules/top-violations | most-triggered rule codes | days, limit, sidekickId |
GET /insights/processing/duration | duration avg + percentiles | days, sidekickId |
GET /insights/processing/time-saved | estimated review time saved | days, sidekickId, manualBaselineSeconds |
/api/integrations.
Parameters
days— size of the rolling window,1–365(default28). Trend series return one point per day in the window, with empty days filled as zero.sidekickId— narrows a metric to a single sidekick. Only present on the endpoints where it applies;by-sidekickandby-file-type-sidekickalready group by sidekick, so they don’t take it.limit— capstop-violations,1–50(default10).manualBaselineSeconds— the manual-review baseline for the time-saved calculation, default300. Time saved per asset ismax(0, manualBaselineSeconds − averageProcessingSeconds).
Example
Notes
- All counts and rates reflect live data for your workspace — no collection or opt-in step is required — and exclude sandbox submissions.
- Processing-duration and time-saved are computed per asset, from the time each asset spends in processing (start to completion). Assets still in flight are excluded until they finish.
- For high-frequency dashboards, cache responses for a short interval — these are daily rollups, so a few seconds of staleness is usually acceptable.