nano SIEM

Introduction to nano

Supported in:Self-hosted

What nano is, how it stores and queries security data, and where to start.

Use nano to ingest security logs, hunt for activity, and build detections and investigations. nano normalizes events to OCSF 1.9 and uses nPL for searches, detections, and dashboards.

What you get

  • One schema. Parsers emit standard OCSF 1.9 events. You query the same dotted paths across every source: src_endpoint.ip, actor.user.name, device.hostname, process.cmd_line, file.hashes.value. See OCSF schema.

  • A piped query language. nPL reads left to right: a search, then commands.

    metadata.log_source=aws_cloudtrail status_id=2 | stats count by actor.user.name, src_endpoint.ip | sort -count
  • Detections that run on a schedule, with staging, bake-in and alerting modes. See Detections.

  • Investigation tools: alerts, cases, notebooks, risk-based alerting and prevalence.

  • pivt, the built-in AI assistant, for drafting parsers, queries, detections and dashboards, and for summarizing investigations.

  • Platform audit as searchable events. Audit records are OCSF API Activity (class_uid=6003) events under metadata.log_source=audit.

Ways to run nano

There are two ways to run nano, both on your own servers. Open (free, AGPL-3.0) covers ingest, parsing, search, detections and alerting. Self-run (paid) adds cases, notebooks, pivt AI, risk analytics, SSO, playbooks, custom enrichments and high availability. If you bought Self-run, start at Self-run.

Who runs itGuide
Docker ComposeYou, on one server, or three or more for HA (Self-run). Open or Self-run.Install with Docker Compose
Kubernetes (Helm)You, on your own cluster. Self-run.Install on Kubernetes (Helm)
Air-gappedYou, in a disconnected network. Self-run.Air-gapped deployment

Every deployment uses the same architecture: Vector for ingest and parsing, ClickHouse for events and PostgreSQL for metadata. Deployment models covers them in detail.

How data flows

  1. A sender delivers raw logs with an ingest token bound to one log source.
  2. That source's parser maps each record to an OCSF event. Fields with no OCSF home are kept under unmapped, and the original line is kept in raw_data.
  3. Events are written to ClickHouse, partitioned by day.
  4. Search, detections and dashboards query them with nPL.

Get started

  1. Install nano on one host, or on Kubernetes.
  2. Send your first feed: create a log source, issue an ingest token and send events.
  3. Learn the schema: which OCSF paths to query for users, hosts, IPs, processes and files.
  4. Search your data: nPL commands and examples.
  5. Write detections: turn a search into a scheduled rule.
  6. Look up a field: OCSF paths by entity, plus the class table.

Have saved queries that use the earlier flat field names? They still resolve for compatibility; see Migrating from legacy field names to move them to OCSF paths.

What's next

On this page