Reasoning Templates API

Pillar 2 endpoints for managing reasoning templates and retrieving rule evaluation results. Reasoning templates define deterministic red-flag rules and verification chains that apply to specific country/vertical combinations. The RedFlagEngine evaluates these rules without any LLM involvement.

Endpoints Summary

Method	Path	Purpose
GET	/api/reasoning-templates	List all reasoning templates
GET	/api/reasoning-templates/{template_id}	Get full template detail
GET	/api/cases/{workflow_id}/rule-evaluations	Get rule evaluation results for a case

---

List Reasoning Templates

GET /api/reasoning-templates

Returns all registered reasoning templates, optionally filtered by country.

Query Parameters:

Name	Type	Required	Description
country	string	No	ISO country code to filter templates (e.g., BE, NL)

Response 200

[
  {
    "id": "be_psp_aml",
    "name": "Belgium PSP AML Template",
    "country": "BE",
    "vertical": "psp_merchant_onboarding",
    "version": "1.0",
    "enabled": true,
    "rule_count": 12,
    "verification_steps": 5,
    "workflow_template_id": "psp_merchant_onboarding",
    "regulatory_framework": "6AMLD"
  }
]

---

Get Template Detail

GET /api/reasoning-templates/{template_id}

Returns the full reasoning template including all red flag rules, verification chain steps, condition types, and action types.

Path Parameters:

Name	Type	Required	Description
template_id	string	Yes	Reasoning template identifier

Response 200

Returns the complete template model including red_flag_rules (with 10 condition types and 5 action types) and verification_chain steps.

Status Codes:

• 200 -- Success
• 404 -- Reasoning template not found

---

Get Rule Evaluations

GET /api/cases/{workflow_id}/rule-evaluations

Retrieves the rule evaluation results for a specific case. These are the outcomes of running the RedFlagEngine against the case's investigation findings.

Path Parameters:

Name	Type	Required	Description
workflow_id	string	Yes	Temporal workflow ID of the case

Query Parameters:

Name	Type	Required	Description
iteration	integer	No	Specific iteration to retrieve (defaults to latest)

Response 200

{
  "evaluated": true,
  "template_id": "be_psp_aml",
  "rules_triggered": 2,
  "confidence_cap": 65.0,
  "evidence_gate": 0.8,
  "results": [
    {
      "rule_id": "shell_company_indicators",
      "triggered": true,
      "condition_type": "finding_category_present",
      "action_type": "cap_confidence",
      "action_value": 65.0,
      "evidence": "Finding 'no_physical_presence' matched category"
    }
  ]
}

When no evaluation has been performed yet, the response returns {"evaluated": false, "results": []}.

Status Codes:

• 200 -- Success (returns empty results if no evaluation exists)