Inforcer Audit Events
Inforcer records an audit trail of activity and changes it observes.
This skill covers searching that trail (auditEvents) and enumerating the
event-type catalog you filter against. It is a read-only history
surface — it tells you what happened and when, which is the evidence
layer behind a drift or posture report.
Read api-patterns first for the gateway
headers, the region requirement, the {success, message, errors, data}
envelope, and continuationToken pagination. Audit searches commonly span
many pages — paging to completion matters more here than almost anywhere
else, because a partial page silently drops events from the window you
think you searched.
Anti-triggers
- The Microsoft 365 unified audit log — this is Inforcer's own
record of what it observed, a much narrower feed than M365's. For
sign-ins, inbox-rule creation, app consents, or admin operations in
the tenant, use
cipp-alerts(cipp_list_audit_logs). - A security detection or alert to triage — an audit event is
history, not a finding. Use
blumira-findings,huntress-incidents, orcipp-alerts. - The current state that drifted — audit answers when and by
whom; the state itself is
inforcer-baseline-alignment.
Tools
inforcer_audit_event_types
List the event-type catalog — the set of event types Inforcer can record. Returns the type identifiers/labels you use to filter a search.
inforcer_audit_event_types()
Pull this first when you need to filter auditEvents by type: it tells you
the valid type values rather than guessing. The catalog is also a useful
map of what kinds of activity Inforcer tracks at all.
inforcer_audit_search
Search the audit event history. Every filter is optional, and they are:
event_types (an array of values from the catalog), date_from /
date_to, page_size, and continuation_token.
inforcer_audit_search(
event_types=["<from the event-type catalog>"],
date_from="2024-02-01",
date_to="2024-02-29"
)
Returns event objects describing what occurred, when, and (where the API exposes it) the actor and target, plus a continuation token for the next page.
There is no tenant filter. The search is account-wide across every
managed tenant — the tool takes no tenant argument, unlike the rest of
this plugin's surface. To answer "what changed on this tenant", filter by
type and window, then attribute each returned event to a tenant from its own
payload and discard the rest. Budget for that: a narrow per-tenant question
still pulls the whole portfolio's events for the window, so keep windows
tight and page to completion before you filter.
What to look for in an audit review
| Pattern | Why it matters |
|---|---|
| Change events that line up with a drift finding | Connects "the tenant drifted" to "here is the change that caused it" |
| Privileged-role or policy changes | Highest blast radius; worth corroborating against identity review |
| A burst of activity around an incident window | Helps reconstruct a timeline for post-incident review |
| No events where you expected some | May indicate a filter mismatch (wrong type or window) rather than genuine quiet |
Workflow patterns
Build a valid filter, then search
types = inforcer_audit_event_types() # discover valid types
events = inforcer_audit_search(
event_types=[types[...]], # values from the catalog
date_from="2024-02-01", date_to="2024-02-29")
mine = [e for e in events if tenant_of(e) == "Acme"] # filter client-side
Discover the event types first so your filter uses real values; an invalid
type quietly returns nothing and looks like "no activity." Page
continuation_token to completion before concluding the window is empty —
and before filtering to a tenant, since the tenant you want may only appear
on a later page.
Drift corroboration
When baseline-alignment shows a tenant has drifted, search the audit trail around the drift window and pick out the events belonging to that tenant to find the change(s) that produced it. Audit gives the narrative behind the alignment delta — useful in a report and in client conversations.
Caveats
- This surface is read-only. You can search and read the audit trail, but you cannot create, edit, or delete audit events, and audit events are history — they don't let you undo or remediate anything.
- The API is community-sourced (no official public docs); the
auditEventsshape, the event-type catalog values, and the exact filter parameter names are illustrative and credited toroyklo/InforcerCommunity. Confirm filter field names against the catalog on first use. - An empty result is more often a wrong filter (an invalid event type or a too-narrow window) than genuine silence — widen and re-check before reporting "no activity." The same goes for a tenant that vanishes from a client-side filter: confirm you paged the whole window first.
- Page to completion. An un-paged audit search is the easiest way to under-report a change window.
Related Skills
- tenant-management - the managed-tenant roster you attribute returned events against
- identity-governance - attribute change events to the identities/roles that made them
- baseline-alignment - the drift the audit trail helps explain
- api-patterns - envelope, continuationToken pagination, region, and the integer-id gotcha