Insights

Automations accumulate. A schedule quietly starts failing, a webhook keeps pointing at a workflow someone deleted, a search index nobody has touched in months keeps sitting there. Every night, Sprigr runs a self-review of your workspace and surfaces anything worth your attention on the Insights page, each finding paired with a suggested fix where a safe one exists.

How the nightly review works

The review runs automatically once per night for every organisation, in your organisation’s own timezone, at a fixed slot between 1am and 6am (each organisation gets its own slot so the nightly cohort is spread out). There is nothing to set up or schedule — it is on by default.

Each run takes a snapshot of your workspace configuration and evaluates a set of checks against it:

Check	What it flags	Severity
Failing schedule	An enabled scheduled task whose most recent run failed — it looks healthy but has silently stopped doing its job.	Warning
Schedule that never fired	A schedule-triggered workflow whose cron has never fired, more than a day after it was created — usually a mis-written cron expression.	Warning
Over-frequent schedule	A schedule-triggered workflow firing more often than every 15 minutes on average.	Info (warning at 5 minutes or less)
Dangling webhook	An enabled webhook endpoint whose destination workflow or agent no longer exists.	Warning
Unbounded search	A scheduled workflow step that pulls from an integration or HTTP source with no result limit and no time window — it re-reads everything on every run.	Info
Unused search index	A search index untouched for 90 days, or never used and at least 90 days old.	Info

Findings are stable across nights: a condition that persists updates the same finding rather than creating a duplicate, and a finding whose underlying condition clears is automatically marked resolved on a later run.

The Insights page

Insights lives in the sidebar under Knowledge & Data, with a badge showing the number of open findings. Everyone in your organisation can open the page and read the findings; changing the autonomy setting and applying fixes are reserved for the owner.

Findings are grouped into Active, Snoozed, and Resolved sections, with active findings sorted most severe first (critical, then warning, then info). Each finding card shows:

A severity indicator and the finding’s title and detail.
A category pill (for example Misconfiguration, Cost, or Unused).
The affected object — the workflow, schedule, or search index — linked so you can jump straight to it.
When the finding was first and last seen.
A Suggested fix line, when a safe fix exists, stating what the fix does, whether it is reversible or destructive, and which autonomy tier it needs.

If nothing is wrong, the page simply says so: “No findings yet.”

Acting on a finding

For an open finding you can:

Fix — apply the suggested fix (owner only). Fixes are conservative, single mutations: disable a failing schedule, retune an over-frequent cron, add a time window to an unbounded search, or delete an unused index. Destructive fixes ask for confirmation first. Some findings (a schedule that never fired, a dangling webhook) are advisory only — there is no single safe mutation, so they carry no fix button and you resolve them at the source.
Snooze — hide the finding for 7 days. It moves to the Snoozed section and comes back if still relevant.
Dismiss — mark it as noise. It moves to the Resolved section and stays out of your way.

Snoozed and dismissed findings can be reopened at any time. Snoozing and dismissing are available to all members; only applying a fix is owner-gated, because a fix mutates workspace objects.

Autonomy tiers

By default the nightly review only reports. The owner can raise how much Sprigr may do about findings on its own, using the Autonomy panel at the top of the Insights page:

Tier	Name	What it does
1	Advisory	Surfaces findings only. You apply fixes yourself. This is the default.
2	Assisted	May apply reversible fixes on its own after the nightly run; destructive ones still need you.
3	Autonomous	May apply fixes on its own.

Destructive fixes — deleting an object — are additionally gated behind a separate Allow destructive fixes opt-in, which only takes effect at the Autonomous tier. So even at tier 3, nothing is deleted unless the owner has explicitly ticked that box; without it, tier 3 behaves like tier 2 for destructive actions.

Auto-applied fixes follow the same rules as manual ones: only open findings are touched (never anything you have snoozed or dismissed), and each fix declares the minimum tier it requires.

Next steps

Schedules & Tasks — The schedules the nightly review keeps an eye on.
Workflows — Schedule-triggered workflows are the subject of several checks.
Inbound Webhooks — Keep webhook destinations current so they do not turn up as findings.