AI Agents

Automated agents for intelligent data model operations

7 agents available
compliance_review Compliance Review scope: model dry-run ~45s
Reviews data model attributes to identify PII, PHI, PCI-DSS relevant data, and other compliance implications. Suggests appropriate compliance tags and data handling requirements.
Agent Parameters
Name Type Required Description
frameworks array No Compliance frameworks to check (GDPR, HIPAA, PCI_DSS, SOX, CCPA)
default: ['GDPR', 'HIPAA', 'PCI_DSS']
min_confidence string
enum: high, medium, low
 No Minimum confidence level to flag
default: medium
include_recommendations boolean No Include handling recommendations
default: True
use_llm boolean No Use LLM for analysis when heuristics are uncertain
default: True
Execute Agent
JSON endpoint: /agents/compliance_review
type_enrichment Type Enrichment scope: model dry-run ~60s
Analyzes attribute names and suggests appropriate data types and business types. Uses pattern matching and AI analysis to detect common patterns like emails, phone numbers, currencies, and dates.
Agent Parameters
Name Type Required Description
only_missing boolean No Only analyze attributes missing types
default: True
include_business_types boolean No Also suggest business types
default: True
min_confidence string
enum: high, medium, low
 No Minimum confidence level to apply changes
default: medium
use_llm boolean No Use LLM for analysis in addition to heuristics
default: True
Execute Agent
JSON endpoint: /agents/type_enrichment
description_enrichment Description Enrichment scope: model dry-run ~30s
Generates clear, professional descriptions for entities and attributes that are missing descriptions. Can also create documentation notes for entities that would benefit from additional context.
Agent Parameters
Name Type Required Description
include_entities boolean No Generate descriptions for entities
default: True
include_attributes boolean No Generate descriptions for attributes
default: True
create_notes boolean No Create documentation notes for entities
default: True
use_llm boolean No Use LLM for generating descriptions
default: True
Execute Agent
JSON endpoint: /agents/description_enrichment
relationship_detection Relationship Detection scope: model dry-run ~30s
Analyzes entity attributes to identify potential foreign key relationships. Uses pattern matching and AI analysis to detect relationships.
Agent Parameters
Name Type Required Description
min_confidence string
enum: high, medium, low
 No Minimum confidence level
default: medium
use_llm boolean No Use LLM for additional detection
default: True
Execute Agent
JSON endpoint: /agents/relationship_detection
llm_health LLM Health Check scope: project ~5s
Verifies LLM connectivity by sending a minimal ping request. Used to ensure AI features are available when users connect to a project.
Execute Agent
JSON endpoint: /agents/llm_health
project_creation Project Creation scope: project dry-run ~120s
Creates initial data models, entities, and attributes based on natural language instructions about industry domain, systems, and intended models. Generates a foundational structure with COMMENT and TODO notes to guide further customization.
Agent Parameters
Name Type Required Description
instructions string Yes Natural language instructions describing the industry domain, systems, and intended data models
create_model boolean No Whether to create a new model (True) or update existing (False)
default: True
model_name string No Optional name for the model (overrides AI suggestion)
Execute Agent
JSON endpoint: /agents/project_creation
summary Change Summary scope: project ~10s
Generates a human-readable summary of changes made to the data model. Uses AI to create natural language descriptions of what was modified, why, and the impact of the changes.
Agent Parameters
Name Type Required Description
session_id string No Session ID to summarize changes from
changes array No List of changes to summarize (if not using session_id)
include_statistics boolean No Include statistics about the changes
default: True
format string
enum: text, markdown, json
 No Output format: text, markdown, json
default: text
Execute Agent
JSON endpoint: /agents/summary
API Usage

Execute agents programmatically via POST /agents/execute:

{
  "agent_key": "type_enrichment",
  "project_id": "your-project-uuid",
  "user_id": "your-user-uuid",
  "model_id": "optional-model-uuid",
  "dry_run": false,
  "user_instructions": "Focus on detecting date and phone number fields",
  "parameters": {
    "use_llm": true
  }
}