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
Commits on main
Lane F: linked occurrences (3.3), scheduled hunts on the lake (5.2), and the outcomes nobody measured (4.6, 9.4) (#1588)
517a08f Beenu Arora committed 17 days ago
Lane D: response that finishes — verb breadth 89 → 165, a bounded model step, time gates, and a ticket-sync defect (#1587)
f61942f Beenu Arora committed 17 days ago
depth2 7.3 + 7.4: agent identity, sub-principal baselines, and the four AI rules they unblock (#1601)
acb398b Beenu Arora committed 17 days ago
feat(1.3): HR and device context connectors, synced on schedule, cited in triage (#1589)
d42b7e5 Beenu Arora committed 17 days ago
Lane A: identity context that changes the verdict — effective permissions, access review, sub-principal baselines, change timeline and identity resolution (#1590)
ba636b4 Beenu Arora committed 18 days ago
depth2 Lane E: land Phase 7 provider coverage, hosted agents and the AI rule pack on main (#1599)
58f42c7 Beenu Arora committed 18 days ago
feat(ingest): anonymizer and hosting classes stamped at ingest (depth2 1.6) (#1586)
4bb84f0 Beenu Arora committed 18 days ago
depth2 Phase 0: kickoff, re-derived baseline and the SIEM-scope decision (#1584)
0b052e5 Beenu Arora committed 18 days ago
Eval honesty wave: serving-model attribution, full-cohort closure accounting, conservative correlation fallback (#1581)
f840cbe Beenu Arora committed 18 days ago
Agentmetry ingest template: AI coding-agent activity and detections (Discussion #1577) (#1580)
f958621 Beenu Arora committed 18 days ago
fix(api): the rail's business-context leg reads the event the writer writes (#1579)
1a9d86d Beenu Arora committed 19 days ago
Asset criticality reaches fusion confidence and action approval, tighten-only (#1578)
df369df Beenu Arora committed 19 days ago
Business context reaches the triage model, and impact statements must cite it (#1575)
828ff9c Beenu Arora committed 19 days ago
feat(escalations): ack-clock escalation bridge for unacknowledged CRITICAL/HIGH cases (#1573)
f8d92e6 alexmateescu committed 19 days ago
fix(deps): clear minimatch/glob ReDoS advisories across all three lockfiles (#1574)
70dd054 Beenu Arora committed 19 days ago
fix(deps): clear the Oct-30 npm advisory wave (follow-redirects, joi, node-forge) (#1572)
1ba1497 Beenu Arora committed 19 days ago
fix(api): SSO activation no longer locks out the admin; funnel lake column (#1569)
013ecd3 Beenu Arora committed 19 days ago
chore(agents): raise auto-triage max_tokens 512 -> 2048 (#1558)
34d6b1d alexmateescu committed 20 days ago
fix(deps): connector-ts SDK on zod ^4.6.6 via zod/v3 compat (GHSA-mrhx-c7xc-m3w5) (#1560)
dff5334 alexmateescu committed 20 days ago
fix(deploy): pass REALTIME_URL and AISOC_INVESTIGATION_MAX_SECONDS through to agents (#1557)
88dc4b8 alexmateescu committed 20 days ago
fix(deploy): mount slack-bot source into teams-bot (fresh-install crash-loop) (#1556)
0297b98 alexmateescu committed 20 days ago
fix(agents): cap prompt bytes at egress; dead-letter deterministic 400s (#1554)
0214735 alexmateescu committed 20 days ago
fix(connectors): exclude scheduler-owned keys on saved-connector test-connection (#1555)
00ea72e alexmateescu committed 20 days ago
fix(web): a self-hosted install's root lands on the console, not the marketing page (#1561)
8ebedd2 Beenu Arora committed 21 days ago
fix(deps): decode-uri-component 0.5.1 for GHSA-qr28-fpjm-wq7f (high, DoS) (#1562)
07417e2 Beenu Arora committed 21 days ago