Webhook Event Catalog

52 event types your org’s webhook endpoints can subscribe to. Every event uses the same envelope (see Webhook Notifications for delivery, signing, and retry mechanics):

{
  "id": "evt-a1b2c3d4",
  "type": "integrity.violation",
  "created_at": "2026-05-09T18:30:00.000Z",
  "account_id": "acct-3f8a1b2c",
  "data": { "...": "event-specific fields, shown per event below" }
}

Subscribing to a new endpoint defaults to event_types: [] (none) — explicit per-event opt-in is required. See Create a webhook endpoint.

Integrity (AIP)

Fired from the gateway’s per-turn integrity analysis.

`integrity.checkpoint`

Fired on every AIP analysis run, regardless of verdict. Subscribe here for a complete audit trail; subscribe to integrity.violation instead if you only want to page on violations.

{ "checkpoint_id": "chkpt-5d7e9f01", "agent_id": "agt-customer-bot-v3", "session_id": "sess-7c8d9e0a", "verdict": "clear", "concerns": [], "reasoning_summary": "..." }

`integrity.violation`

Fired when AIP analysis returns a boundary_violation verdict. A companion integrity.checkpoint always fires for the same checkpoint — this event is a filtered subset of it.

{ "checkpoint_id": "chkpt-5d7e9f01", "agent_id": "agt-customer-bot-v3", "session_id": "sess-7c8d9e0a", "verdict": "boundary_violation", "concerns": ["..."], "reasoning_summary": "..." }

`conscience.escalation`

Fired when AIP analysis detects a conscience-value conflict in agent reasoning beyond a routine boundary check.

{ "checkpoint_id": "chkpt-xyz", "agent_id": "agt-x", "session_id": "sess-y", "conflicts": [{ "rule": "respects_user_consent", "severity": "high" }], "verdict": "boundary_violation" }

`conscience.values_updated`

Fired when an org owner/admin creates, updates, or removes an org-level conscience value.

{ "org_id": "org-acme", "action": "create", "value_id": "cv-respects-consent", "actor": "user-owner-42" }

Alignment traces (AAP)

Fired when an AAP alignment trace is verified.

`trace.created`

{ "trace_id": "trc-abc", "agent_id": "agt-x", "session_id": "sess-y" }

`trace.verified`

Fired when a trace passes verification (signature, chain, and boundary checks).

{ "trace_id": "trc-abc", "agent_id": "agt-x", "verification_id": "ver-def" }

`trace.failed`

Fired when a trace fails verification — signature mismatch, chain break, or boundary violation.

{ "trace_id": "trc-abc", "agent_id": "agt-x", "failure_reason": "signature_mismatch" }

`trace.escalation_required`

{ "trace_id": "trc-abc", "agent_id": "agt-x", "escalation_reason": "boundary_violation_with_attestation_chain_anomaly", "severity": "critical" }

`policy.violation`

Fired when policy evaluation flags a violation in agent behavior.

{ "agent_id": "agt-x", "policy_id": "policy-no-external-write", "violation_type": "forbidden_tool", "session_id": "sess-y" }

Drift & team coherence

Cross-turn and cross-team detector firings. These are operator-actionable signals — the agent itself never sees them.

`sideband.drift.fired`

Fired when N consecutive non-clear AIP verdicts in a session cross the configured drift threshold.

{ "alert_id": "drift-9a8b7c6d", "agent_id": "agt-customer-bot-v3", "session_id": "sess-7c8d9e0a", "severity": "warning", "drift_direction": "worsening", "sustained_checks": 5, "message": "Detected 5 consecutive non-clear verdicts in session sess-7c8d9e0a" }

`drift.resolved`

Fired when an org owner/admin acknowledges a previously-fired drift alert. Companion to sideband.drift.fired.

{ "alert_id": "drift-9a8b7c6d", "org_id": "org-acme", "acknowledged_by": "user-owner-42" }

`sideband.coherence.fired`

Fired when a team’s pairwise governance floor drops below threshold, conflict edges exceed threshold, or specific agents’ declared values diverge from the rest of the team.

{ "team_id": "team-research-pod", "reasons": ["pairwise_governance_floor below threshold (0.42 < 0.5)"], "pairwise_governance_floor": 0.42, "conflict_edge_count": 4, "outlier_agent_ids": ["agt-rogue-experimenter"], "severity": "warn" }

`sideband.fault_line.fired`

Fired when team fault-line analysis detects a minority of agents whose alignment cards diverge from the team majority.

{ "team_id": "team-research-pod", "fault_lines": [{ "id": "fl-1", "value": "respects_user_consent", "severity": "high", "affected_agents": ["agt-rogue"] }], "severity": "high", "affected_agent_ids": ["agt-rogue"] }

`sideband.fleet.fired`

Fired when team-topology analysis detects outlier agents, a cluster partition, or a pair score below threshold across a team.

{ "team_id": "team-research-pod", "pattern_type": "cluster_partition", "severity": "high", "pattern_detail": { "components": [["agt-x", "agt-y"], ["agt-rogue"]] } }

Safe House

Fired by the Safe House front door and the review workflow. See Safe House Threat Model for what each verdict means. An sh.evaluation.* event is emitted per screened surface, not per turn — one request can emit several (the inbound message, plus each tool result screened before the model sees it).

`sh.evaluation.warn`

{ "evaluation_id": "eval-abc", "agent_id": "agt-x", "session_id": "sess-y", "verdict": "warn", "reasons": ["recipe_match_low_confidence"] }

`sh.evaluation.quarantine`

{ "evaluation_id": "eval-abc", "agent_id": "agt-x", "session_id": "sess-y", "verdict": "quarantine", "quarantine_id": "q-xyz", "reasons": ["pii_detected"] }

`sh.evaluation.block`

{ "evaluation_id": "eval-abc", "agent_id": "agt-x", "session_id": "sess-y", "verdict": "block", "reasons": ["forbidden_tool_attempt"] }

`sh.canary.triggered`

Fired when a planted canary credential appears in agent output or input — a strong signal of active exploitation.

{ "canary_id": "cnry-7e", "agent_id": "agt-x", "session_id": "sess-y", "surface": "outgoing" }

`sh.session.escalated`

Fired when a session’s accumulated risk score crosses an escalation threshold.

{ "session_id": "sess-y", "agent_id": "agt-x", "session_risk_score": 0.85, "threshold": 0.75 }

`sh.campaign.detected`

Fired when cross-session correlation detects a coordinated attack targeting your agents.

{ "campaign_id": "camp-9f", "indicators": ["repeated_canary_attempt", "clustered_session_escalation"], "affected_agents": ["agt-x", "agt-y"] }

`sh.review.triggered`

Fired when a quarantine-band verdict raises a review hold — the conversation is held pending adjudication.

{ "review_id": "rev-abc123", "agent_id": "agt-x", "session_id": "sess-y", "evaluation_id": "eval-abc", "surface": "tool_calls", "source_verdict": "quarantine", "sla_deadline": "2026-07-09T19:30:00.000Z" }

`sh.review.approved`

Fired when a reviewer releases a held item — it proceeds.

{ "review_id": "rev-abc123", "agent_id": "agt-x", "session_id": "sess-y", "reviewer_kind": "builtin_opus", "reason": "Tool call reads a public config value; no protected-surface overlap on inspection." }

`sh.review.denied`

Fired when a reviewer denies a held item — a transparent refusal is applied in its place.

{ "review_id": "rev-def456", "agent_id": "agt-x", "session_id": "sess-y", "reviewer_kind": "builtin_opus", "reason": "Tool call arguments target a protected asset; withholding." }

`sh.review.expired`

Fired when a review hold reaches its SLA deadline unresolved. The card’s configured on_timeout default applies (fail-closed reject unless configured otherwise).

{ "review_id": "rev-ghi789", "agent_id": "agt-x", "session_id": "sess-y", "on_timeout_applied": "reject", "sla_deadline": "2026-07-09T19:30:00.000Z" }

Detection reports

`recipe.candidate.created`

Fired when your org submits a false-negative/false-positive report on a detection recipe (or Mnemom’s own arena tooling proposes one). Delivered to your org’s own webhook endpoints when you’re the reporting org, alongside Mnemom’s internal review queue.

{ "candidate_id": "cand-abc12345", "writer_identity": "customer-fn-report", "report_type": "fn", "related_recipe_id": "rcp_0123456789abcdef", "agent_id": "agt-x", "checkpoint_id": "chk-xyz", "summary_excerpt": "Agent was prompted to leak credentials via Base64-encoded payload; existing prompt-injection rule didn't fire." }

The recipe lifecycle events that follow a candidate’s triage — recipe.promoted, recipe.retired, advisory.published, ioc.added, network.campaign.closed, and reviewer-mode.changed — are internal Mnemom operations events today (delivered only to Mnemom’s own account, never to a customer org), so they aren’t in this catalog. Poll Intelligence Overview for advisories and indicators instead.

Agent lifecycle

`agent.paused` / `agent.resumed` / `agent.killed`

Fired when an org owner/admin (or auto-containment) changes an agent’s containment status. killed is stronger than paused — reactivation requires explicit operator action.

{ "agent_id": "agt-x", "org_id": "org-acme", "action": "pause", "actor": "user-admin", "reason": "Investigating drift incident", "previous_status": "active", "new_status": "paused" }

`agent.exemption.granted` / `agent.exemption.revoked`

Fired when an org admin grants or revokes a temporary exemption from a Safe House section for a specific agent.

{ "agent_id": "agt-x", "org_id": "org-acme", "exemption_id": "exempt-abc", "exempt_section": "back_door.outgoing.pii", "exempt_patterns": ["customer_id"], "reason": "Customer-id sharing required for support flow", "granted_by": "user-admin", "expires_at": "2026-06-09T00:00:00Z" }

Governance (cards & templates)

`alignment_card.updated` / `protection_card.updated`

Fired when an agent’s alignment or protection card is updated.

{ "agent_id": "agt-x", "card_id": "card-agt-x-alignment-v7", "actor": "user-admin", "change_summary": "Tightened integrity_mode from observe to nudge" }

`org_alignment_template.updated` / `org_alignment_template.deleted`

Fired when an org-scope alignment template is set, updated, or cleared — this triggers a recompose for every agent inheriting from the template.

{ "org_id": "org-acme", "template_id": "tpl-org-alignment-v2", "actor": "user-owner", "agents_flagged_for_recompose": 47 }

`org_protection_template.updated` / `org_protection_template.deleted`

Same shape as the alignment-template events above, for the protection template.

Teams

`team.created` / `team.archived`

{ "team_id": "team-research-pod", "name": "Research Pod", "org_id": "org-acme" }

`team.member_added` / `team.member_removed`

{ "team_id": "team-research-pod", "agent_id": "agt-x" }

`team.card_updated`

Fired when a team’s alignment- or protection-template card is set, updated, or cleared.

{ "team_id": "team-research-pod", "card_id": "card-team-research-protection-v3" }

Reputation

`reputation.score_changed`

Fired when an agent’s reputation score crosses a meaningful threshold.

{ "agent_id": "agt-x", "previous_score": 0.82, "new_score": 0.71, "delta": -0.11, "reason": "boundary_violation_streak" }

`reputation.grade_changed`

Fired when an agent crosses a reputation-grade boundary.

{ "agent_id": "agt-x", "previous_grade": "A", "new_grade": "B", "reason": "score_drop_below_0.75" }

Quota & billing

`quota.warning` / `quota.exceeded`

Fired once per billing period when integrity-check usage crosses 80% / 100% of your usage budget.

{ "usage_percent": 85, "checks_used": 8500, "checks_included": 10000 }

`quota.risk_warning` / `quota.risk_exceeded`

Same shape, for risk-assessment usage.

{ "usage_percent": 105, "risk_assessments_used": 525, "risk_assessments_included": 500 }

`quota.team_reputation_warning` / `quota.team_reputation_exceeded`

Same shape, for team-reputation computation usage.

{ "usage_percent": 110, "team_reputations_used": 110, "team_reputations_included": 100 }

`quota.sh_warning`

Fired when Safe House usage cost for the period crosses your configured budget alert threshold.

{ "current_cost_cents": 5500, "threshold_cents": 5000, "sh_checks_used": 1100, "sh_checks_included": 1000 }

`subscription.status_changed`

Fired when your billing subscription transitions state (e.g. active → past_due).

{ "status": "past_due", "previous_plan_id": "plan-team", "new_plan_id": "plan-team", "cancel_at_period_end": false }

`transaction.completed`

Fired when a billed autonomous-operation transaction completes.

{ "transaction_id": "tx-7a8b9c", "agent_id": "agt-customer-bot", "team_id": "team-research", "action_type": "autonomous_operation", "duration_seconds": 4.2 }

​Integrity (AIP)

​integrity.checkpoint

​integrity.violation

​conscience.escalation

​conscience.values_updated

​Alignment traces (AAP)

​trace.created

​trace.verified

​trace.failed

​trace.escalation_required

​policy.violation

​Drift & team coherence

​sideband.drift.fired

​drift.resolved

​sideband.coherence.fired

​sideband.fault_line.fired

​sideband.fleet.fired

​Safe House

​sh.evaluation.warn

​sh.evaluation.quarantine

​sh.evaluation.block

​sh.canary.triggered

​sh.session.escalated

​sh.campaign.detected

​sh.review.triggered

​sh.review.approved

​sh.review.denied

​sh.review.expired

​Detection reports

​recipe.candidate.created

​Agent lifecycle

​agent.paused / agent.resumed / agent.killed

​agent.exemption.granted / agent.exemption.revoked

​Governance (cards & templates)

​alignment_card.updated / protection_card.updated

​org_alignment_template.updated / org_alignment_template.deleted

​org_protection_template.updated / org_protection_template.deleted

​Teams

​team.created / team.archived

​team.member_added / team.member_removed

​team.card_updated

​Reputation

​reputation.score_changed

​reputation.grade_changed

​Quota & billing

​quota.warning / quota.exceeded

​quota.risk_warning / quota.risk_exceeded

​quota.team_reputation_warning / quota.team_reputation_exceeded

​quota.sh_warning

​subscription.status_changed

​transaction.completed

​See also

Integrity (AIP)

`integrity.checkpoint`

`integrity.violation`

`conscience.escalation`

`conscience.values_updated`

Alignment traces (AAP)

`trace.created`

`trace.verified`

`trace.failed`

`trace.escalation_required`

`policy.violation`

Drift & team coherence

`sideband.drift.fired`

`drift.resolved`

`sideband.coherence.fired`

`sideband.fault_line.fired`

`sideband.fleet.fired`

Safe House

`sh.evaluation.warn`

`sh.evaluation.quarantine`

`sh.evaluation.block`

`sh.canary.triggered`

`sh.session.escalated`

`sh.campaign.detected`

`sh.review.triggered`

`sh.review.approved`

`sh.review.denied`

`sh.review.expired`

Detection reports

`recipe.candidate.created`

Agent lifecycle

`agent.paused` / `agent.resumed` / `agent.killed`

`agent.exemption.granted` / `agent.exemption.revoked`

Governance (cards & templates)

`alignment_card.updated` / `protection_card.updated`

`org_alignment_template.updated` / `org_alignment_template.deleted`

`org_protection_template.updated` / `org_protection_template.deleted`

Teams

`team.created` / `team.archived`

`team.member_added` / `team.member_removed`

`team.card_updated`

Reputation

`reputation.score_changed`

`reputation.grade_changed`

Quota & billing

`quota.warning` / `quota.exceeded`

`quota.risk_warning` / `quota.risk_exceeded`

`quota.team_reputation_warning` / `quota.team_reputation_exceeded`

`quota.sh_warning`

`subscription.status_changed`

`transaction.completed`

See also