Skip to main content

Use A Local LLM For Private Reports

These retained guides document AdversaryGraph v7.0.0. Latest published stable release: v8.1.2. Current development: 8.1.2, verified 2026-10-20 at source commit 789d874. Historical screenshots and instructions are not evidence of current-release behavior.

Level: Intermediate

Goal: Analyze sensitive content without public LLM routing.

Real-Life Scenario​

A customer report contains sensitive incident details, so the analyst must run extraction through a private local LLM gateway instead of a public API.

When To Use This​

Use this workflow when you need a structured analyst workflow and want the output to remain traceable to evidence.

Steps​

  1. Configure local/private LLM gateway in deployment env.
  2. Run selftest and confirm provider is reachable.
  3. Choose local provider in AI Analysis.
  4. Analyze the report and review mappings.
  5. Export only reviewed findings.

Expected Result​

Private report extraction with controlled model routing.

Review Notes​

  • Keep source labels and evidence attached to every accepted result.
  • Treat actor matches, enrichment hits, and matrix overlap as analytical signals until corroborated.
  • Export only reviewed findings for customer, SOC, detection engineering, or executive use.