Skip to main content

AiSOC

Open-source AI SOC platform. Agent decisions are recorded in an investigation ledger and a public eval harness runs in CI. MIT-licensed and self-hostable.

How AiSOC compares

AiSOC is open-source and self-hostable, and every agent decision is recorded in the investigation ledger. Closed-source AI SOCs run on vendor infrastructure and do not expose the agent loop, which makes them harder to review under SOC 2, ISO 27001, or DORA controls.

CapabilityAiSOCWazuhSplunk Enterprise SecurityClosed-source AI SOC
Open-source (MIT) and self-hostableYes — MITYes — GPLv2NoCloud-only
Agent decisions are auditable line-by-lineYes — full ledger + replayNo agent layerBlack-box MLBlack-box agent
Substrate has a public regression-gate harness200-case suite, CI-gatedNot publishedNot publishedVendor-claimed only
Native AI investigation agentLangGraph multi-agentNoSplunk AI Assistant add-onClosed-source
MITRE ATT&CK heatmap + purple-team emulationBuilt-inPartialPremium add-onLimited
Plugin SDK (Python + Go) + community marketplaceBoth SDKs, MITWodles onlySplunkbaseVendor-only
Compliance evidence (SOC2 / ISO / NIST / DORA)Built-in dashboardsNoPremium add-onReporting only

Capability claims for other vendors are sourced from their public documentation as of 2026. AiSOC's claims map directly to code in this repository — see the docs.

What ships in the box

The first three rows describe the properties most relevant to a regulated buyer. The remainder enumerate the SOC substrate.

Auditable agent decisions

Every prompt, tool call, and rationale the agent emits is persisted to the investigation ledger and replayable step-by-step in the case workspace.

Public eval harness

200-incident, CI-gated regression harness over the AiSOC substrate (extractors, fusion, templates, judges). Reproducible on a laptop in seconds. The page describes what each metric measures and what it does not.

MIT-licensed, self-hostable

The code, prompts, and templates are in the repo. No CLA, no telemetry, no calls home.

LangGraph multi-agent investigation

Recon, forensic, responder, and reporter agents wired through a LangGraph orchestrator for triage and case enrichment.

Playbook engine

Visual React Flow editor with 12 starter templates for automated, human-gated response actions.

UEBA

Per-user Welford online baselines, Z-score anomaly scoring, and Kafka-integrated anomaly publishing.

Honeytokens

HMAC-SHA256 signed deceptive credentials (URL, file, AWS key, email) with first-touch webhook alerting.

Purple Team

Atomic Red Team YAML parser, Caldera executor, ATT&CK coverage heatmap, and tabletop sessions.

Real-time fusion

Kafka spine with sub-second alert ingestion, Bloom-filter dedup on 10M+ IOCs, ML scoring (LightGBM + Isolation Forest).

Attack graph

Neo4j entity graph with attack-path reconstruction and blast-radius gating on automated actions.

Detection engineering

Sigma over OpenSearch and ClickHouse, YARA, KQL/EQL — community catalog with one-click install.

Enterprise governance

SAML 2.0 and OIDC SSO, multi-tenant Postgres RLS, granular RBAC, and immutable audit log.

Compliance dashboards

SOC 2, ISO 27001, NIST CSF, PCI-DSS, HIPAA, and DORA evidence with MTTD/MTTR/MTTC SLA tracking.

Plugin ecosystem

Python and TypeScript SDKs, Ed25519-signed publishing, and a community marketplace.

Deployment

Helm charts, Docker Compose, OpenTelemetry traces/metrics/logs, and PostgreSQL backup with KMS encryption.