Events
An event is something Merge Pilot announces: a merge without approval, a change approval waiting for a decision, secrets found in a merge request, a connection that stopped working. Each notification route picks the events it takes (see Routes), and some events are also emailed to the people they concern (see Personal notifications). This page lists every event.
How to read the list
- Event is the label of the route's checkbox on the Notifications tab and of the switch in My profile → Notifications. The checkbox's tooltip shows the API value, which is also the
eventof a webhook delivery and the value in a route'seventslist. Every event (all) takes all of them, including events added later. - Sent when says what raises the event. Sent once per says what makes it new: the same event about the same thing in the same state is sent once, however often Merge Pilot notices it, and checks running at the same moment send it once too. Where a time is given (within an hour), the same event is sent again only after that time.
- Title and Message are what Slack, Teams and email show.
MR !12/PR #12andMR/PRfollow the project's VCS (GitLab merge requests, GitHub pull requests);group/appstands for the project's path (repo <id> when the repository is not registered in Merge Pilot);KEY-1for the change approval's key in the tracker. Titles are one line of at most 120 characters, messages at most 2,000 (see Message content). - Colour is the bar or title colour: green (good), amber (warning) or red (danger).
- Opens is the page the message's Open in Merge Pilot link goes to.
- Personally to is who gets the event by email besides the routes (see Personal notifications). Every personal email is on until the person turns it off, except Connection recovered, which is off until they turn it on. A dash: routes only. Nobody is emailed personally about their own action (an admin who turns a connection off gets no email about it; the other admins do). The routes still get the event.
Nothing is sent for some merge requests and repositories whatever the routes say: see What holds events back.
Merge requests
| Event (label in the UI) | API value | Sent when | Sent once per | Title | Message | Colour | Opens | Personally to |
|---|---|---|---|---|---|---|---|---|
| Merge request evaluated | mr_processed | A standard change passed the checks of its policy and its merge gate turned approved. Other change types do not send it | Head commit | MR approved (standard change) | MR !12 (group/app) passed the checks of policy Name and may merge | Green | The merge request | — |
| Merge request blocked | mr_blocked | The merge gate holds a merge request because a blocking control failed, and its policy does not close it. Not for imported past merge requests | Head commit and set of failing controls | MR blocked; MR would be blocked in a watched repository | MR !12 (group/app) cannot merge under policy Name: <why>; in a watched repository … would be blocked under policy Name: <why>. Its repository is watched, so nothing was written | Red | The merge request | The author |
| Merge request refused | mr_rejected | A policy rejected and closed the merge request (Close the merge request, see Merge gate) | Head commit | MR rejected | MR !12 (group/app) was rejected and closed: <reason> | Red | The merge request | The author |
| Not enough approvers | mr_approvers_short | Fewer people may approve the change than its policy needs | Head commit | Not enough approvers | MR !12 (group/app) needs 2 approvals under policy Name, but only 1 person may approve it (… but nobody may approve it) | Red | The merge request | The author |
| Risk assessed | mr_risk_assessed | A head commit was assessed high or critical risk (see Risk). The new score at merge time sends nothing | Head commit | MR high risk, or MR blocked for its risk when the risk blocks the merge | MR !12 (group/app) was flagged, risk level high (or was blocked, risk level critical) | Amber; red when blocked | The merge request | — |
| Merge request merged | mr_merged | A merge request was merged | Merge request | MR merged | MR !12 (group/app) was merged by alice | Green | The merge request | — |
| Merged without approval | mr_merged_without_approval | A merge request under change management was merged without an approved change approval (also one merged before Merge Pilot evaluated it) | Merge request | MR merged without approval | MR !12 (group/app) was merged by alice without change-management approval | Red | The merge request | — |
| Merged during a change freeze | mr_merged_in_freeze | The change was merged during a change freeze that held it (see Change calendar) | Merge request and freeze | Merged during a change freeze | MR !12 (group/app) was merged by alice during the change freeze Name, which held it | Red | The merge request | — |
| Separation of duties broken | mr_sod_violation | Someone who approved the change merged it, and the policy does not allow it (see Separation of duties) | Merge request and approver | Separation of duties broken | MR !12 (group/app): alice approved the change and merged it, which its policy does not allow | Red | The merge request | — |
| Approval revoked after merge | mr_approval_revoked_after_merge | The change approval of a merged change was withdrawn in the tracker (see Change approval lifecycle) | Merge request and change approval | Approval revoked after merge | MR !12 (group/app) is merged, but the approval of change approval KEY-1 was withdrawn in the tracker | Red | The change approval | — |
| Emergency declared | emergency_declared | A change was declared an emergency (see Emergency changes) | Merge request, policy and who declared it | Emergency change declared | MR !12 (group/app) was declared an emergency by bob (policy Name): it may merge before it is approved (its author when nobody else is named) | Amber | The merge request | Company admins |
| Emergency review overdue | emergency_review_overdue | The review of an emergency change was not decided by its deadline (see Review deadline) | Change approval | Emergency change review overdue | The review of emergency change MR !12 (group/app) in change approval KEY-1 is past its deadline | Red | The change approval | Company admins |
Change approvals
| Event (label in the UI) | API value | Sent when | Sent once per | Title | Message | Colour | Opens | Personally to |
|---|---|---|---|---|---|---|---|---|
| Change approval created | change_approval_created | A change approval was opened for a merge request (see Change approval lifecycle) | Change approval | Change approval created | Change approval KEY-1 was created for MR !12 (group/app) | Green | The change approval | The assignee; when nobody was assigned, everyone who may decide it |
| Change approval approved | change_approval_closed | A change approval was approved, in the tracker or in Merge Pilot | Change approval and decision round: approved again after its approval was revoked, or after it was reopened for new commits, it is sent again | Change approval approved | Change approval KEY-1 was approved, so MR !12 (group/app) is approved | Green | The change approval | The author |
| Change approval rejected | change_approval_rejected | A change approval was rejected, in the tracker or in Merge Pilot | Change approval and decision round | Change approval rejected: MR closed when the policy closes the merge request, else Change approval rejected: MR blocked | Change approval KEY-1 was rejected, and its policy closes MR !12 (group/app), or Change approval KEY-1 was rejected: MR !12 (group/app) stays open and cannot merge | Red | The change approval | The author |
| Change approval reassigned | change_approval_reassigned | Someone took over a change approval from its assignee (Take over, see Voting), or a change approval nobody could take was assigned once someone could | The same handover, within an hour | Change approval reassigned | bob took over change approval KEY-1 of MR !12 (group/app) from alice, or Change approval KEY-1 of MR !12 (group/app) is now assigned to bob | Amber | The change approval | The new assignee |
| Change approval due soon | change_approval_due_soon | A change approval's deadline is near (a quarter of its time, at most a day, see Change approval deadlines): a normal change's change approval deadline, or an emergency change's review deadline | Change approval and deadline | Change approval due soon | Change approval KEY-1 of MR !12 (group/app) is due on 2026-10-30 14:00 UTC; for an emergency change The review of emergency change MR !12 (group/app) in change approval KEY-1 is due on … | Amber | The change approval | The assignee |
| Change approval overdue | change_approval_overdue | A normal change's change approval is past its deadline and still not decided | Change approval and deadline | Change approval overdue | Change approval KEY-1 of MR !12 (group/app) was due on 2026-10-30 14:00 UTC and is still not decided | Red | The change approval | The assignee |
| Approval fatigue | approval_fatigue | An approver's decision raised an approval concentration warning (see Approval fatigue) | Approver, within the company's lookback days | Approval concentration warning | MR !12 (group/app): alice approved 25 change approvals in the last 7 days | Amber | The merge request | — |
Opens: the change approval is the issue in the built-in tracker when the change approval is there, and the merge request otherwise (an issue in Jira is not a page of Merge Pilot).
Change approval deadlines
A normal change's change approval can have a deadline: Change approval deadline (hours) on the company's Approvals tab, which a policy's approval control can replace with its own Change approval deadline (see Approval → Change approval deadline). The deadline is set when the change approval is created. Emergency changes keep their policy's review deadline.
A change approval that is opened again gets a new deadline, counted from that moment with the same hours: reopened for new commits, or its approval revoked in the tracker. What was sent about the old deadline does not hold back the new one's notices.
Change approval due soon is sent when a quarter of the time to the deadline is left, and at most a day before it: a day before a deadline of four days or more, 6 hours before a 24-hour deadline, 30 minutes before a two-hour one. Once the deadline passes and the change approval is still open, Change approval overdue is sent (for an emergency change, Emergency review overdue). Merge Pilot checks every few minutes; each is sent once per deadline, and an audit entry is written with it.
Scan findings
| Event (label in the UI) | API value | Sent when | Sent once per | Title | Message | Colour | Opens | Personally to |
|---|---|---|---|---|---|---|---|---|
| Secrets found in a merge request | secrets_detected | The secret scan of a merge request found potential secrets (see Secret scan) | Head commit and number of findings | Secrets found in MR | MR !12 (group/app): 3 potential secrets found in the diff | Red | The merge request | The author |
| Vulnerable dependency introduced by a merge request | vulnerabilities_detected | The dependency scan of a merge request found vulnerabilities it introduces (see Dependency scan) | Head commit, number and worst severity | Vulnerable dependencies in MR | MR !12 (group/app) introduces 2 dependency vulnerabilities (1 critical) | Red when one is critical, amber otherwise | The merge request | The author |
| Malicious package in a merge request | malicious_package_detected | A merge request adds packages known to be malicious | Head commit and number | Malicious package in MR | MR !12 (group/app) adds 1 package known to be malicious: remove them | Red | The merge request | The author |
| Denied license in a merge request | license_denied | A merge request adds dependencies under a license the company denies (see Licenses) | Head commit and number | Denied license in MR | MR !12 (group/app) adds 2 dependencies under a license the company denies | Red | The merge request | The author |
| Dependency source problems in a merge request | dependency_sources_detected | A merge request's dependency files take packages from sources the dependency source check fails (see Dependency sources) | Head commit and number | Dependency sources to check in MR | MR !12 (group/app): 2 dependency source problems in the dependency files | Amber | The merge request | The author |
| CI configuration problems in a merge request | ci_config_problems | The CI configuration a merge request changes has problems (see CI configuration check) | Head commit and number | CI configuration problems in MR | MR !12 (group/app): 2 problems in the CI configuration it changes | Amber | The merge request | The author |
| Vulnerable container images in a merge request | image_vulnerabilities_detected | The container image scan of a merge request found vulnerable packages in the images it adds or changes (see Container image scan) | Head commit, number and worst severity | Vulnerable container images in MR | MR !12 (group/app): 5 vulnerabilities in 2 container images it refers to (1 critical) | Red when one is critical, amber otherwise | The merge request | The author |
| Infrastructure problems in a merge request | iac_problems_detected | The infrastructure check of a merge request found misconfigurations it brings into the files it changes; those the target branch already has are left out (see Infrastructure check) | Head commit, number and worst severity | Infrastructure problems in MR | MR !12 (group/app): 3 problems in the infrastructure files it changes (worst: high) | Red when the worst is critical, amber otherwise | The merge request | The author |
| New vulnerability on a default branch | vulnerability_found | A round of checks found new vulnerabilities on default branches, at or above the company's threshold (see Vulnerability inventory). The baseline and findings an exception covers are not announced | One digest per round of checks | New vulnerabilities | New vulnerabilities on the default branch of group/app: (or … on the default branches of 3 projects:) then one line per finding, worst first, at most 10: • <advisory> <package>@<version> (<severity>), with in <project> when there are several projects, , fixed in <version> and , introduced by MR !12 when known; +N more for the rest | Red when the worst is high or critical, amber otherwise | Insights → Vulnerabilities | — |
Each scan finding is its own event: a merge request with secrets and a denied license sends both, and each is sent without the others.
New vulnerability on a default branch collects the findings of every project in a round into one message per route. A route with path patterns gets only the projects it matches; a route with Stop here keeps the projects it took from the routes after it. A round still waiting for its last checks after an hour sends what it has.
Repositories
| Event (label in the UI) | API value | Sent when | Sent once per | Title | Message | Colour | Opens | Personally to |
|---|---|---|---|---|---|---|---|---|
| Direct change to a watched branch | direct_change | A change reached a watched branch without a merge request, and the company announces direct changes. Not for ignored accounts. See Direct changes → Notifications | Change: the branch moving on from the head Merge Pilot last saw (a push delivered again, or found by both the push and the head check, is sent once). A branch deleted again, restored and deleted again, or force-pushed back, is a change of its own and is sent and recorded again. The first change to a branch is sent at once; further ones to it within the next 15 minutes are sent together, in one notification, when the 15 minutes end | Direct change to <branch>; for the 15 minutes' further changes, N more direct changes to <branch> | alice pushed 2 commits to main of group/app without a merge request; alice force-pushed main of group/app, rewriting its history (2 commits not from a merge request); alice deleted the branch main of group/app; then . The branch is under change management: this is a violation or . Change management would have blocked it. Someone (found by the head check) when the pusher is not known. Further changes: 2 more direct changes were made to main of group/app without a merge request within 15 minutes of the first one (3 commits not from a merge request), by alice, bob. Its head is now 1a2b3c4d (or The last one deleted the branch), then . The branch is under change management: these are violations or . Change management would have blocked them | Red | The project | Company admins |
| Branch protection weakened | protection_weakened | The branch protection check found that a repository under change management lets changes skip the merge gate, or found other gaps than last time, and the company announces direct changes (see Repositories settings). Fixing the gaps sends nothing; when gaps come back, it is sent again | Project and set of gaps | Branch protection weakened | Changes can reach the target branches of group/app without change management: then one line per gap | Red | The project | Company admins |
| API token expiring | api_token_expiring | A personal API token expires within 14 days, then within 3 days (see API tokens) | Token and warning | API token <name> expires in <N> days | The API token deploy (mpt_ab12_…) expires on 2026-12-01 UTC and cannot be extended. Create a new token under My profile → API tokens, put it where this one is used, then delete this one | Amber | My profile | Its owner |
| Service API token expiring | service_token_expiring | A service API token expires within 14 days, then within 3 days | Token and warning | API token <name> expires in <N> days | …Create a new token under the company's API tokens… | Amber | The company's API tokens tab | Company admins |
| Connection token expiring | connection_token_expiring | The token a GitLab connection uses expires within 14 days, then within 3 days | Connection, expiry and warning | Token of <connection> expires in <N> days | The token of the GitLab connection GitLab expires on 2026-12-01. Then Merge Pilot can no longer reach the VCS: create a new token on GitLab for the same account and enter it on the connection | Amber (red within 3 days) | The company's Connections tab | Company admins |
| Change approvals of a deactivated person | change_approval_assignee_deactivated | A person deactivated in Merge Pilot was assigned open change approvals (see Approvals → When an assignee is deactivated) | Person and set of change approvals | Change approvals of <person>, deactivated | alice was deactivated while assigned 2 open change approvals (CMG-12, CMG-14): they are assigned again by their policies' rules (or …: they stay assigned to alice until someone takes them over) | Amber | The merge requests | Company admins |
| Protection removed from a governed branch | branch_unprotected_by_admin | A company admin removed the protection of a branch under change management through Merge Pilot (see Branch protection) | Every removal | Protection removed from <branch> | alice removed the protection of main in group/app, a branch under change management. Changes can now reach it without the merge gate | Red | The project | Company admins |
Connections and sign-in
| Event (label in the UI) | API value | Sent when | Sent once per | Title | Message | Colour | Opens | Personally to |
|---|---|---|---|---|---|---|---|---|
| Connection problem | connection_unhealthy | A connection got a problem it did not have: the VCS refuses Merge Pilot's credentials, the GitHub App (or an installation) lacks permissions Merge Pilot needs, the VCS's API rate limit is used up, or an admin turned the connection off. See Connection health and How connection health is decided | Problem, within an hour; an admin turning the connection off, every time | Connection problem: <connection> | The GitLab connection Name has a problem: the VCS refused Merge Pilot's credentials (or … the GitHub App lacks permissions Merge Pilot needs, or an installation has not accepted them, … the VCS's API rate limit is used up: background work waits until it resets, … it was turned off) | Red | The company's Connections tab | Company admins |
| Connection recovered | connection_recovered | The connection's problem is gone, or it was turned on again, and that problem was announced | Announced problem; turned on again, every time | Connection recovered: <connection> | The GitLab connection Name works again. The problem was: …, or The GitLab connection Name was turned on again | Green | The company's Connections tab | Company admins, off until they turn it on |
| Identity provider unreachable | sso_check_failed | When a session was refreshed, Merge Pilot could not ask the identity provider whether the person may still sign in; the person is asked to sign in again (see Identity providers) | Identity provider, within an hour | Identity provider unreachable: <provider> | Merge Pilot could not ask the identity provider Name whether people may still sign in, so they are asked to sign in again | Red | The company's Sign-in tab | Company admins |
| LDAP directory unreachable | ldap_check_failed | The same for the company's LDAP directory (see LDAP directory) | Within an hour | Directory unreachable | Merge Pilot could not ask the company's LDAP directory whether people may still sign in, so they are asked to sign in again | Red | The company's Sign-in tab | Company admins |
How connection health is decided
Merge Pilot checks every connection regularly, and when an admin clicks Test on it. A problem is cleared only by a check that uses the same credentials that failed:
- GitLab: the check uses the connection's token, the one all work uses. A working check clears credentials refused and rate limit used up.
- GitHub App: the check first calls GitHub as the App, then with each installation's token, the credentials all work uses (a call that does not use up the rate limit). A refused installation token keeps the problem even while the App itself answers, until every installation Merge Pilot can use works again. Suspended installations, and installations removed on GitHub, do not count: they show on the connection (see GitHub App installations). An installation whose rate limit is used up shows rate limit used up. Permissions missing is decided from the permissions last read from GitHub, by the regular check and by Test alike.
- Rate limit hit by a background job: when the VCS gave a reset time (its Retry-After, or the limit's reset), the problem stays until that time has passed, even if a check works: a check cannot see every limit a job can hit. A limit a check finds itself is looked at again by the next check.
- Rate limit kept for live events: when only the share of the hour's calls kept for webhooks and people's actions is left, background work waits. Rate limit used up stays until the limit resets, even if a check works.
The same problem is announced once within an hour: a connection that keeps failing and recovering does not flood the routes. A recovery is announced only when its problem was: a problem held back by the hour leaves its recovery unannounced too, so the announcements always alternate. When nothing of a connection event could be queued, the change is taken back and made again by the next check. The connection's state on the Connections tab is always current.
People and the platform
| Event (label in the UI) | API value | Sent when | Sent once per | Title | Message | Colour | Opens | Personally to |
|---|---|---|---|---|---|---|---|---|
| Role changed | user_role_changed | A person's role changed, by an admin, or by their identity provider's or company directory's groups at sign-in | Every change | Role changed: <user> | alice's role changed from user to policy_manager (by bob) (or (by the identity provider Name), (by the company directory)) | Amber | The Users page | The person whose role changed |
| VCS access to review | access_released | Merge Pilot stopped tracking access it gave on the VCS, or tokens it created, which may still work there: a person was deactivated, moved to another company, made platform admin or unlinked their VCS account, a project was removed, a connection deleted or the company deactivated (see People) | Every release | VCS access to review | Merge Pilot stopped tracking 2 access grants and 1 token of alice because the person was deactivated. They may still work on the VCS: review them under the company's people | Amber | The company's People tab | Company admins |
| License expiring | license_expiring | Merge Pilot's license expires in 30, 7 or 1 days, and once more when it has expired and its grace period starts (see Platform settings). Sent to every active company | Expiry date and threshold | License expires in 7 days; when the grace period starts, License expired: grace period until 2026-11-20 | Merge Pilot's license expires on 2026-11-06 00:00 UTC. A platform admin renews it under Settings; when the grace period starts, Merge Pilot's license expired on … UTC. It keeps working until … UTC, then stops taking events from the VCS and trackers. A platform admin renews it under Settings | Amber; red at 7 days or fewer, and when the grace period starts | No link | Company admins |
| Background job failed | job_failed | A background job of the company failed its last attempt (see Jobs). A failure that shows a connection problem is sent as Connection problem instead; a change approval that keeps failing to open for the same reason is reported once | VCS connection and kind of failure, within an hour | Background job failed | The secret scan job <job ID> for MR !12 (group/app) failed after its retries: the credentials were refused. The full error is in the audit log | Red | Insights → Jobs | — |
| Notification not delivered | notification_failed | A delivery to a route failed for good (see Delivery log and failures). Never sent to the route that failed, and not for a failed route test or a failed personal email | Route and kind of failure, within an hour | Notification not delivered: <route> | A secrets_detected notification could not be delivered to the route Name: the endpoint answered 500. Its deliveries are in the delivery log | Red | The company's Notifications tab | Company admins |
Background job failed says what kind of failure it was (the credentials were refused, permission denied, rate limited, TLS certificate problem, timed out, a server could not be reached, not found, a server answered with an error, or an internal error), never the error itself: errors can name internal hosts and addresses. The full error is in the audit log (job_failed).
What holds events back
Before any route or person is looked at:
| What | What it holds back |
|---|---|
| Imported merge requests | Events about merge requests imported from the history are never sent |
| Merge requests left watched | A merge request recorded while its repository was watched stays watched after the repository moved to change management, and never sends events |
| Watched repositories (What is announced, off by default) | While it is off, events about a watched repository or a watched merge request are not sent to any route or person, webhooks included. See What is announced |
| Direct changes (What is announced, on by default) | While it is off, Direct change to a watched branch and Branch protection weakened are not sent |
| New vulnerabilities on default branches (What is announced, High and above by default) | The lowest severity that New vulnerability on a default branch announces; Never sends none |
Some events are never sent for watched merge requests whatever the settings, because nothing is written for them: Merge request evaluated, Emergency declared, and Merge request refused (a watched merge request is never closed). A watched merge that change management would have held, or that falls in a change freeze, is recorded on the merge request as evidence and not sent. See Watching repositories.
Related pages
- Routes: which events go to which channel.
- Personal notifications: who gets which event by email, and how to turn it off.
- Channels: how the messages look in Slack, Teams and email.
- Outbound webhooks: the same events as JSON.
- Audit log: most events are also written there.