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
-- Migration 041: SOC console parity — alert correlation narrative
-- Implements W6 (structured InvestigationRail) from the v1.5 SOC Console
-- Parity plan.
--
-- Changes:
-- alerts.narrative (TEXT, nullable) — a deterministic, human-readable
-- correlation narrative cached on the alert. Generated by the fusion
-- service at correlate-time and refreshed by the API at read-time when
-- missing or stale. The InvestigationRail surfaces this verbatim so the
-- right pane renders instantly without an LLM round-trip; the
-- on-demand `ExplainDrawer` (renamed "Deep Explain" in the rail) still
-- provides the streaming LLM explanation when an analyst wants more.
--
-- Backfill: leave narrative NULL on existing rows. The API layer will
-- compute and cache narrative lazily on first read of an alert, so
-- legacy rows pick it up the moment an analyst opens them.
--
-- Idempotent — safe to re-run.
BEGIN;
ALTER TABLE alerts
ADD COLUMN IF NOT EXISTS narrative TEXT;
COMMENT ON COLUMN alerts.narrative IS
'Deterministic correlation narrative cached at fusion time. '
'NULL means "not yet computed"; the API will lazily compute and '
'persist on first read of the alert. Surfaced verbatim by the '
'InvestigationRail. For the streaming LLM explanation, see the '
'Deep Explain action which goes through /api/v1/alerts/{id}/explain.';
COMMIT;