Skip to main content

Dashboard

The Dashboard is the first page after signing in. It answers what is waiting on someone now, and how has change gone over the last weeks?: the open merge requests that cannot move until someone acts, and a set of figures about merge requests, risk, scans, jobs and activity.

For a short overview of what it is for, see the Insights feature page.

Opening it​

Dashboard in the side menu (the first entry). Every role sees it; Recent activity only those who may read the audit log.

WhoWhose data
Company admin, policy manager, auditor, userTheir own company. The page says Change activity at <company>.
Platform admin, with a company selected in the headerThe selected company
Platform admin, no company selectedEvery company together: Change activity across all companies. The leaderboards, Change types and Security summary then stay empty (they are per company)

What is counted​

Unless a card says otherwise:

  • Watched and change-managed merge requests are both counted. Merge requests of watched repositories, and those imported from a watched repository's history, are in the counts. The exceptions are named below (Needs attention, MRs Blocked by SCA, Cycle time breakdown).
  • "Last 30 days" is a rolling period ending now, not calendar months.
  • When a merge request was opened is when it was opened on the VCS, also for a merge request imported from a watched repository's history. Only when the VCS did not say is it when Merge Pilot first recorded it. The cycle time breakdown, the risk card, the leaderboards and the security summary are the exceptions: they count from when Merge Pilot first recorded the merge request.

Needs attention​

Open changes that cannot move until someone acts. The panel lists the open merge requests that are:

  • refused by their policy;
  • held by an enforced check that failed, or by a risk hold;
  • not a standard change, and their approval failed (for example, the change approval was rejected);
  • an emergency change whose review is not done yet;
  • not a standard change, with a change approval that waits for a decision.

Watched merge requests never appear: nothing holds them. These are the merge requests of the Needs attention filter of the merge request list (see Filters and search).

At most 8 rows are shown: first those that are stopped, then those on a hold, then those waiting on a person, newest first within each. Each row shows the title, the merge request's number and author (!12 by alice), its clearance track in short, and when it was opened. Select a row to open the merge request.

ShownWhen
View all NMore than 8 merge requests need attention. Opens the merge request list filtered to Needs attention
View merge requests8 or fewer. Opens the same list
Nothing is waiting (Every open merge request is either cleared or moving through change management on its own.)None needs attention
Merge requests could not be loaded, with a retry buttonThe list could not be read

Key figures​

Figure (label in the UI)What it countsLine below it
Open merge requestsMerge requests that are open now<time> avg merge: the average time from when a merge request was opened to its merge, over the merge requests merged in the last 30 days. A merge request whose opening the VCS did not date, and that Merge Pilot recorded only after it merged, is left out
Merged, last 30 daysMerge requests merged in the last 30 daysN closed without merging: those closed without merging in the last 30 days
Failed jobs, last 24 hoursBackground jobs that failed and finished in the last 24 hours (shown in red when there are any)N total jobs: every job kept, in any state
Active tracker issuesIssues of the built-in tracker (not archived) whose state is in the To do or In progress category, as the issue's own tracker project defines that state (two projects may give one state name different categories). A state of no known category counts as To doN done: those in a Done state

The average merge time is shown as <1m, minutes (42m) or hours and minutes (27h 5m).

MR Throughput (30 days)​

An area chart of the last 30 days (days in UTC, today included):

  • Opened: merge requests opened that day;
  • Merged: merge requests merged that day.

Risk, last 30 days​

How the merge requests first recorded in the last 30 days that have a risk assessment are spread over the risk levels Low, Medium, High and Critical: a bar of their shares, then each level with its count and percentage, and N assessed merge requests. Levels without any merge request are left out.

Without any assessment the card says No risk assessments yet (Risk is assessed when a project has risk scoring enabled and a merge request is processed.). See Risk.

Cycle time breakdown​

How long merged merge requests took, phase by phase.

Window: 7d, 30d (the default) or 90d: the merge requests merged in that many days up to now that a policy matched. Merge requests imported from a watched repository's history are left out: they never went through Merge Pilot, so their phases are unknown. Watched merge requests are included.

Phase (label in the UI)FromTo
Policy matchedMerge Pilot first recorded the merge requestA policy matched it
Risk assessmentThe policy matchedThe risk was assessed (zero when it was not)
Approval waitThe risk assessment (or the policy match)The merge request was approved (zero when no approval was recorded)
MergeThe approval (or the step before)The merge

The total runs from when Merge Pilot first recorded the merge request to its merge.

  • Percentile Breakdown (N MRs): a stacked bar for P50, P75 and P90: each segment is that phase's own percentile. N is the number of merge requests counted.
  • Weekly P50 Trend: the median total time per week (weeks start on Monday, in UTC), by the week of the merge. With fewer than two weeks it says Not enough data for trend.
  • N skewed (warning sign): merge requests whose timestamps are out of order, so a phase came out negative. The negative phase is counted as zero.
  • Averages (information sign): percentiles could not be computed, so averages are shown for every percentile.
  • No merged MRs in this period when nothing was merged in the window.

Top authors, Top projects, Top approvers​

Three lists of the top five over the last 30 days, each with its count and a bar relative to the first:

List (label in the UI)Counts
Top authorsMerge requests first recorded in the last 30 days, by author (their VCS username)
Top projectsMerge requests first recorded in the last 30 days, by repository (its name in Merge Pilot)
Top approversMerge requests approved in the last 30 days, by who approved them

A list without data says No data.

Change types​

The number of merge requests whose policy matched in the last 30 days (changes matched a policy in the last 30 days), a bar of their shares, and the count of each change type: standard, normal and emergency. Selecting a count opens the merge request list filtered by that change type (that list is not limited to 30 days).

Security summary​

Over the merge requests first recorded in the last 30 days (MRs opened in last 30 days):

Figure (label in the UI)Counts the merge requests
MRs with Secretswhose secret scan found at least one possible secret
MRs with SCA Vulnswhose dependency scan found at least one vulnerability the merge request brings in
MRs Blocked by SCAwhose enforced dependency scan failed. Watched merge requests do not count: nothing blocks them

See Secret scan and Dependency scan.

Job queue​

The jobs kept for the company, by state: Pending, Processing (claimed by a worker or running), Completed, Failed and Retry. See Jobs.

Recent activity​

The 15 newest entries of the company's audit log: who, the action, the resource, how long ago, and the result (success, failure). Without entries: No recent activity. See Audit log.

Only those who may read the audit log see this card: platform admins, company admins and auditors. Policy managers and users do not; the page leaves the card out.

Refreshing​

  • While the browser tab is visible, Needs attention, the key figures, the charts, the risk card, the job queue, recent activity and the cycle time breakdown are read again every minute. The leaderboards, Change types and Security summary are read again every 10 minutes. A hidden tab is not refreshed; it refreshes once when it is shown again.
  • Refresh reads the key figures, the charts, the risk card, the job queue and recent activity again at once.
  • The server keeps the figures for a short time: the same company's figures may be up to 15 seconds old, the leaderboards, Change types and Security summary up to 5 minutes old.
  • A first load that fails shows the error with a retry button. A later refresh that fails keeps the figures on screen and says, next to Refresh, The latest refresh failed (…); the figures below may be out of date.

API​

Every endpoint answers for the caller's own company. Platform admins add company_id; without it they get every company together.

RequestAnswers
GET /api/v1/dashboardmr_stats (total_open, merged_30d, closed_30d, blocked, reviews_due, reviews_overdue, avg_merge_time_ms, secrets_detected), job_stats (total, pending, processing, completed, failed, retry, failed_last_24h), tracker_stats (total_open, total_in_progress, total_done), risk_breakdown (low, medium, high, critical), mr_throughput[] (date, opened, merged) and recent_events[] (action, resource, resource_id, username, status, created_at, details; null for a policy manager or user)
GET /api/v1/dashboard/statstop_authors[] and top_approvers[] (name, count), top_projects[] (repo_id, name, count), change_types (standard, normal, emergency, total) and security_summary (secrets_found_mrs, sca_vuln_mrs, sca_blocked_mrs, total_scanned_mrs). Empty without a company
GET /api/v1/dashboard/cycle-time with window (7d, 30d, 90d; 30d by default)breakdown (policy_match, risk_assessment, approval_wait, merge and total, each with p50, p75, p90 in milliseconds; sample_size, bad_data_count), trend[] (week_start, total_p50, match_p50, risk_p50, approval_p50, merge_p50, count), window and approximate (averages instead of percentiles)
GET /api/v1/merge-requests?attention=trueThe merge requests of Needs attention (see Merge requests → API)

Fields the page does not show:

  • blocked: open merge requests whose merge gate is blocked, watched ones left out.
  • reviews_due: emergency changes with a review deadline whose change approval is not approved, rejected or cancelled; reviews_overdue: those past the deadline.
  • secrets_detected: merge requests first recorded in the last 30 days with at least one possible secret.
  • total_scanned_mrs: merge requests first recorded in the last 30 days that have a secret scan or a dependency scan.

Reading the Dashboard is not recorded in the audit log.