Skip to content

EntiHub Documentation

AI Agent Identity Registry

AI Agent Identity Registry

Treat each AI agent as a governed enterprise asset—not a chat experiment. EntiHub stores agent identity, ownership, allowed data domains, risk classification, and approval status so platform, security, and business owners share one register.

Why this is needed

Most companies cannot answer which agents exist, who owns them, or what data they may read.
Audits and AI governance need maker/checker approval and history of agent changes.
Agents and CI pipelines need a machine-readable source of truth via API/CLI.

Proposed entities in EntiHub

Governance settings

Setting	Recommended value	Why
`approvalRequired`	`true`	Agent go-live and access changes must be approved.
`historyEnabled`	`true`	Recertification and incident forensics need full history.
`allowSameAuthorApproval`	`false`	Maker/checker separation for sensitive changes.
`deletedRetentionDays`	`90` (or higher)	Soft-delete retention for auditability.

Entity definition (YAML sketch)

name: ai_agent
displayName: AI Agent
domain: AI Governance
description: AI Agent
version: "1.0.0"
approvalRequired: true
historyEnabled: true
deletedRetentionDays: 90
displayRefColumn: code
uniqueConstraints:
  - name: uq_ai_agent_code
    columns: [code]
consumption:
  views:
    - schema: mdm_consume
      name: vw_ai_agents_approved
      apiName: ai-agents-approved
      where: "[status] = 'approved'"
columns:
  - name: code
    dataType: string
    nullable: false
    maxLength: 64
  - name: name
    dataType: string
    nullable: false
    maxLength: 128
  - name: owner_team
    dataType: string
    nullable: false
    maxLength: 128
  - name: purpose
    dataType: string
    nullable: false
    maxLength: 1000
  - name: data_domains
    dataType: string
    nullable: false
    maxLength: 512
  - name: risk_level
    dataType: string
    nullable: false
    maxLength: 16
  - name: status
    dataType: string
    nullable: false
    maxLength: 16
  - name: model_name
    dataType: string
    nullable: true
    maxLength: 64
  - name: prompt_version
    dataType: string
    nullable: true
    maxLength: 32
  - name: recertify_by
    dataType: date
    nullable: true

YAML sketches

Sketches highlight the important fields and governance flags. Trim or extend columns to match your domain. See Setup Entity for the full YAML reference.

Sample data

code	name	owner_team	purpose	data_domains	risk_level	status	model_name	prompt_version	recertify_by
SALES_FORECAST_AGENT	Sales Forecast Agent	Sales Analytics	Monthly sales forecast briefs	sales,customers,products	medium	approved	gpt-4.1	7	2027-01-15
AP_INVOICE_TRIAGE	AP Invoice Triage	Finance Ops	Classify invoice exceptions	suppliers,invoices	high	approved	gpt-4.1	3	2026-12-01
HR_POLICY_BOT	HR Policy Bot	People Ops	Answer policy questions	hr_policies	low	draft	gpt-4.1-mini	1
SUPPLIER_RISK_AGENT	Supplier Risk Agent	Procurement	Flag high-risk supplier changes	suppliers,payments	critical	suspended	gpt-4.1	5	2026-10-01

Benefits for the company

Single inventory of agents for security, compliance, and platform teams.
Clear ownership and approved data domains before production use.
Re-certification dates drive periodic review.
API/CLI consumers can refuse unapproved or suspended agents.