Open-source AI Security Operations Center: alert fusion, LLM-agent triage, MITRE ATT&CK investigation, and a replayable decision ledger for every agent step. Self-hostable, runs with no API keys, MIT licensed. Ships an MCP server for Claude, Cursor and Continue.
Python agentic-aiai-securityalert-triageblue-teamcybersecuritydetection-engineeringincident-responselanggraphllm-agentsmcpmitre-attackmodel-context-protocolpurple-teamsecurity-operationsself-hostedsiemsoarsocthreat-detectionthreat-intelligence
/**
* Hand-authored TypeScript types mirroring the AiSOC OpenAPI schema.
*
* A curated, ergonomic subset intended for direct use by SDK consumers — not
* a generated file. This header used to say the types were kept in sync with
* docs/openapi.yaml "via the `pnpm codegen` script which regenerates
* src/openapi.d.ts". There was no such file, nothing in CI ran that script,
* and the sentence was the only stated reason to trust these declarations.
*
* What is actually enforced is narrower and true: `scripts/check_sdk_surface.py`
* proves every endpoint `client.ts` calls is an operation the spec declares.
* The field-level shapes below are still hand-kept.
*/
// ── Enums ─────────────────────────────────────────────────────────────────────
export type AlertSeverity = "critical" | "high" | "medium" | "low" | "info";
export type AlertStatus = "open" | "in_progress" | "closed" | "false_positive";
export type CasePriority = "critical" | "high" | "medium" | "low";
export type CaseStatus = "open" | "investigating" | "resolved" | "closed";
// ── Core models ───────────────────────────────────────────────────────────────
export interface Alert {
id: string;
tenantId: string;
title: string;
severity: AlertSeverity;
status: AlertStatus;
source: string;
sourceRef?: string;
mitreTactics: string[];
aiScore?: number;
caseId?: string;
createdAt: string;
updatedAt: string;
}
export interface Case {
id: string;
tenantId: string;
caseNumber: string;
title: string;
status: CaseStatus;
priority: CasePriority;
assignee?: string;
mitreTactics: string[];
alertIds: string[];
createdAt: string;
updatedAt: string;
}
export interface DetectionRule {
id: string;
tenantId: string;
name: string;
description?: string;
ruleLanguage: string;
severity: AlertSeverity;
enabled: boolean;
createdAt: string;
updatedAt: string;
}
export interface Connector {
id: string;
tenantId: string;
name: string;
connectorType: string;
isEnabled: boolean;
healthStatus: string;
eventsIngested: number;
createdAt: string;
updatedAt: string;
}
export interface Playbook {
id: string;
name: string;
description?: string;
version: string;
steps: PlaybookStep[];
triggerConditions?: Record<string, unknown>;
createdAt: string;
updatedAt: string;
}
export interface PlaybookStep {
id: string;
name: string;
type: string;
action?: string;
parameters?: Record<string, unknown>;
nextSteps?: string[];
}
export interface PlaybookRun {
runId: string;
playbookId: string;
status: string;
startedAt: string;
completedAt?: string;
triggerData?: Record<string, unknown>;
stepResults?: Record<string, unknown>;
}
export interface ApiKey {
id: string;
name: string;
prefix: string;
scopes: string[];
expiresAt?: string;
lastUsedAt?: string;
createdAt: string;
}
// ── Request / response envelopes ─────────────────────────────────────────────
export interface Page<T> {
items: T[];
total: number;
page: number;
pageSize: number;
}
export interface ErrorResponse {
detail: string;
code?: string;
}
// ── Filter / pagination params ────────────────────────────────────────────────
export interface PaginationParams {
page?: number;
pageSize?: number;
}
export interface AlertFilters extends PaginationParams {
severity?: AlertSeverity;
status?: AlertStatus;
caseId?: string;
search?: string;
}
export interface CaseFilters extends PaginationParams {
status?: CaseStatus;
priority?: CasePriority;
assignee?: string;
}
export interface ApiKeyCreateRequest {
name: string;
scopes: string[];
expiresAt?: string;
}
export interface ApiKeyCreateResponse {
key: ApiKey;
/** Raw key — only returned on creation, store it safely. */
rawKey: string;
}
// ─── Responder ───────────────────────────────────────────────────────────────
export type ApprovalStatus = "pending" | "approved" | "denied" | "expired";
export type ApprovalRisk = "low" | "medium" | "high" | "critical";
/** An action an agent proposed that needs a human decision before it runs. */
export interface Approval {
id: string;
tenant_id: string;
run_id: string | null;
case_id: string | null;
alert_id: string | null;
requested_by: string;
required_user_id: string | null;
required_topic: string | null;
title: string;
summary: string;
risk_level: ApprovalRisk;
/**
* The action itself: `action_type`, `target`, `parameters`. After a
* decision it also carries a `dispatch` record saying whether the action
* actually executed — the approval row is the answer to "was this done",
* not just "was this decided".
*/
action: Record<string, unknown>;
status: ApprovalStatus;
decided_by_id: string | null;
decided_at: string | null;
decision_comment: string | null;
expires_at: string | null;
created_at: string;
updated_at: string;
}
export interface ApprovalFilters extends PaginationParams {
status?: ApprovalStatus;
/** Only approvals routed specifically to the calling user. */
mine?: boolean;
risk_level?: ApprovalRisk;
}
/** A W3C PushSubscription, as `PushSubscription.toJSON()` produces it. */
export interface PushSubscriptionPayload {
endpoint: string;
keys: { p256dh: string; auth: string };
/** Topics to receive: `p0_alert`, `agent_approval`, `oncall_handoff`. */
topics?: string[];
}
export interface OnCallEntry {
user_id: string;
name: string;
email: string | null;
starts_at: string;
ends_at: string;
rotation: string | null;
}
/** One registered live action: a vendor, a verb, and its contract. */
export interface LiveActionDescriptor {
vendor_id: string;
capability: string;
description?: string;
[key: string]: unknown;
}
export interface LiveActionDiscovery {
executors: LiveActionDescriptor[];
vendors: string[];
capabilities: string[];
}