nano SIEM
Agents & MCP

pivt

Supported in:Self-hosted

nano's built-in AI assistant: the @ command panel, AI wizards for detections, tuning, dashboards, parsers and queries, and background case investigation.

Use pivt to draft queries, detections, dashboards, and parsers and to investigate your data. You can work in the chat panel, open an AI wizard, or let background case investigation run.

pivt appears in three ways:

  • The pivt panel, where you chat or type @ commands.
  • AI wizards inside the pages where you already work: rule editor, tuning, dashboards, parser editor and search.
  • Background AI that works without being asked: automated case investigation.

pivt is included with nano Self-run. The open-core build doesn't include it; there, use a coding agent with the nanodac and Investigator MCP servers instead.

The pivt panel

Type free text to chat. The first message starts a notebook for the conversation, and later messages continue it.

Type @ at the start of the input, or after a space, to run a command:

CommandWhat it does
@search <query>Opens Search with the query
@pivot <query>Opens Search with the query
@entity <value>Opens Search for the entity
@detection <text>Opens the rules list filtered by the text
@case <text>Opens the cases list filtered by the text
@summarizeOpens Analyze on the current search results
@tuneIn the rule editor, opens AI tuning for the rule; elsewhere, opens AI Tuning
@dashboard <prompt>Opens the dashboard generation wizard with your prompt
@playbook <prompt>Opens a new playbook with the drafting wizard
@note <text>Adds a note to the notebook
@pin <text>Pins a note
@handoff <text>Records a handoff to a collaborator
@evidence <text>Adds an evidence note
@runbookNot available yet

There are no slash commands. See Notebook commands for how entries appear in a notebook.

AI wizards

PageWhat pivt does
Rules: new ruleDrafts a detection (nPL, MITRE mapping, triage hints) from a description
Rules: triage hintsSuggests ignore_when / suspicious_when hints
AI TuningReviews a rule's matches and proposes tuning changes
DashboardsBuilds a dashboard from a prompt; the panel editor turns a question into nPL
Add data and Log SourcesGenerates or edits a parser from sample logs
SearchBuilds nPL from a question, corrects a failing query, reviews a query
Search: AnalyzeSummarizes the current results
NotebooksSummarizes a notebook, analyzes a note, suggests queries, builds a timeline
PlaybooksDrafts a response playbook
Marketplace: custom enrichmentGenerates custom enrichment code

The wizards are independent of Cases: they work on searches, rules, dashboards and parsers directly.

Background AI

pivt also works in the background:

  • Case investigation. When a case is created, alerts are grouped into a case, or alerts are added to it, pivt runs an investigation in the case notebook. It extracts entities and records searches for each one (pivots, activity roll-up, source overview, timeline sample), then hunts further and writes a narrative summary. If AI can't run, the deterministic searches are still recorded and the summary says why the narrative was skipped.

The ai search command

| ai sends the result rows of a search to pivt for classification and adds ai_verdict, ai_confidence and ai_reasoning to each row:

class_uid=1007 process.name="powershell.exe" | head 50 | ai prompt="Is this command line likely malicious?" max_rows=50

At most 500 rows are sent. A slow model call doesn't fail the search. On open core, every row gets ai_verdict="SKIPPED".

Configure the provider

You bring your own AI provider and choose models under Settings → AI & pivt. See AI & pivt.

Troubleshoot

Need help with these checks? Contact support with the request ID and time with timezone.

AI stays off

Check that an AI provider is configured and enabled under Settings → AI & pivt. See AI & pivt.

What's next

On this page