Schema reference

Every entry in this knowledge base follows a structured schema — version 0.2. This page documents the schema for contributors, automation developers, and anyone building on the knowledge base content.

Entry structure

Each entry is a Markdown file with YAML frontmatter and structured content sections.

Frontmatter fields

id: string                    # Unique identifier — e.g. c2-prompt-injection
title: string                 # Full entry title
sidebar_label: string         # Short label for navigation
tags: list[string]            # Searchable tags
last_verified: date           # ISO date of last fact-check pass
next_review: date             # ISO date triggering automated review

Layer structure

Layer	Heading	Audience	Key fields
1	## Layer 1 — Start here	All audiences	Headline, summary, severity badge, key question, persona tabs
2	## Layer 2 — Practitioner overview	Risk, compliance, PMs	Risk description, likelihood drivers, consequence types, controls summary table
3	## Layer 3 — Controls detail	Risk practitioners, audit	Full control descriptions, jurisdiction notes, KPIs
4	## Layer 4 — Technical implementation	Engineers, security analysts	Code examples, tool references, compliance implementation

Persona tabs (Layer 1)

Each entry includes persona-specific hooks in a tabbed interface. All entries include three practitioner tabs. Entries linked from the Fork everyday track (A1, C4, E1) also include an Everyday tab written in plain language for general public readers.

<Tabs>
  <TabItem value="executive" label="Executive / Board">
  [2-3 sentences anchored to reading an audit report or board paper]
  </TabItem>
  <TabItem value="pm" label="Project Manager">
  [2-3 sentences anchored to a go-live sign-off decision]
  </TabItem>
  <TabItem value="analyst" label="Security Analyst">
  [2-3 sentences anchored to implementing controls or reviewing a vendor]
  </TabItem>
  {/* Everyday tab — only on A1, C4, E1 (Fork-linked entries) */}
  <TabItem value="everyday" label="Everyday">
  [Plain language for general public. No jargon, no framework codes.
   Ends with a link back to the Fork scenario. See CONTENT_STYLE_GUIDE for voice rules.]
  </TabItem>
</Tabs>

Controls summary table (Layer 2)

Every control in the controls summary must have these five fields:

Field	Values	Notes
Control	string	Short descriptive name
Owner	function name	Security, Technology, Risk, Legal, HR, Compliance, Procurement, Operations, All
Effort	Low / Medium / High	Low = <1 week, Medium = 1–4 weeks, High = >1 month
Go-live?	Required / Post-launch badge	required-golive or required-post CSS class
Definition of done	string	One observable, verifiable sentence

Severity badges

<span className="severity-badge severity-critical">Critical severity</span>
<span className="severity-badge severity-high">High severity</span>
<span className="severity-badge severity-medium">Medium severity</span>
<span className="severity-badge severity-low">Low severity</span>

Severity defaults are starting points — real assessments are context-dependent.

Framework chips

<span className="framework-chip">NIST AI 600-1</span>
<span className="framework-chip">EU AI Act Art. 15</span>

Scenario seed (end of entry)

Every entry includes a scenario seed for training module use:

**Context:** [Organisational setting]
**Trigger event:** [What happens to initiate the scenario]
**Complicating factor:** [What makes resolution non-obvious]
**Discussion questions:** [3-5 questions]
**Difficulty:** Foundational | Intermediate | Advanced
**Jurisdictions:** [Applicable jurisdictions]

Jurisdictional scope

Entries include jurisdiction notes on controls where obligations differ. The controlled vocabulary for jurisdictions is: AU, EU, US, UK, Global.

Global means the risk and controls apply without jurisdiction-specific framing.

Verify flags

Claims that require verification before publication are flagged inline: [VERIFY: description]. The automation engine tracks these and clears them after a successful verification pass.

Version history

Version	Changes
v0.1	Initial schema — four layers, incidents, framework refs, scenario seed
v0.2	Added jurisdictional scope fields, controls_summary with owner/effort/done, persona hooks in Layer 1

Full template

The complete entry template is available in the repository: schema/entry_template.md