Polyodos — Guide utilisateur

English

Français

English

Français

Appearance

✦ AI Scoping ​

AI Scoping applies to IEC 62443 the same principle as the AI assistant of the other modules (EBIOS RM, Industrial Risks, EVRP): describe the OT/ICS context in natural language and let an AI engine pre-fill part of the analysis — here the Context, the initial ZCR2 assessment and the ZCR3 zones/conduits.

The AI engine used is configurable at /gestion/ia among the 8 available providers (Gemini, Claude, Mistral, Groq, HuggingFace, OpenRouter, Ollama, Custom) — see the AIConfigBanner component at the top of the page to switch providers without leaving the screen.

Two sections ​

A navigation bar offers: 📖 Usage guide and ✦ Scoping sheet.

📖 Usage guide ​

Recaps the place of AI Scoping in the module's AI roadmap:

  1. Scoping (this page) → feeds Context, ZCR2, ZCR3
  2. ZCR5 — threat scenario generation (coming soon)

A "Point-in-time help" panel details the module's other AI entry points: CVE suggestion (ZCR1), per-zone SL-T suggestion (ZCR2), threat scenario generation (ZCR5), gap analysis (Compliance).

✦ Scoping sheet ​

  • Installation description sheet — a text area where you paste a Markdown description of the SUC (process, equipment, network, exposure). Ready-to-use example buttons: Chemical industry example, Drinking water example, Energy example, and a Clear button. A character counter indicates whether the sheet is rich enough (> 500 characters).
  • Images (optional) — a drop zone (drag-and-drop or click), PNG/JPG/WEBP formats, 3 images maximum. A warning appears if the selected AI engine doesn't support image analysis.
  • Sections to pre-fill — 3 pre-selected checkboxes: IEC 62443 Context, ZCR2 — Initial Risk Assessment, ZCR3 — Zones & Conduits.
  • ✦ Analyze the sheet and pre-fill the IEC 62443 analysis button — disabled if the sheet is under 100 characters, if no API key is configured, or if no section is checked.

Irreversible action

A confirmation dialog warns that analyzing the sheet erases the existing data of the checked sections before repopulating them. Save or export beforehand if the project already contains work on these sections.

Result ​

After analysis, a results block displays: the number of elements imported by type, direct navigation links to Context / ZCR2 / ZCR3, and — for superadmin accounts — the raw JSON returned by the AI engine. On error, a collapsible "Raw response" detail helps with diagnosis.

What the AI generates ​

From the sheet (and any images), the AI engine produces a structured JSON covering:

  • Context — general fields (title, SUC, scope…)
  • ZCR2 — initial risk scenarios (asset, feared event, type, likelihood/severity bounded from 1 to 4)
  • ZCR3 — zones (zone type, target level SL1-SL4) and conduits (linked by zone name)

Adapted to the OT/ICS context

Unlike the AI scoping of the other modules, this one reasons in terms of security zones and Security Levels (SL-T) rather than generic supporting assets/feared events — the description sheet must therefore describe the industrial process and its network segmentation to be usable.

Polyodos — Documentation utilisateur