formulation-orchestrator

name: formulation-orchestrator version: 1.0.0 description: Orchestrates formulation workflow for Amb Wellness/TDS author: AI Orchestrator Team created: 2026-02-04

triggers:

• formulate
• formulation
• create formula
• batch selection
• select batches
• formular
• nueva formula
• formulacion

metadata: raven: emoji: "🧪" priority: 80 category: formulation

capabilities: - batch_selection - tds_compliance - cost_calculation - optimization - report_generation

dependencies: - formulation_reader - frappe

Formulation Orchestrator Skill

Overview

The Formulation Orchestrator is the main coordinator for formulation workflows in the Raven AI Agent system. It manages the complete process from batch selection through report generation.

Architecture

FormulationOrchestratorSkill (main skill)
    ├── BatchSelectorAgent (Phase 2)
    ├── TDSComplianceAgent (Phase 3)
    ├── CostCalculatorAgent (Phase 4)
    ├── OptimizationEngine (Phase 5)
    └── ReportGenerator (Phase 6)

Workflow Phases

Phase 1: Request Analysis

• Parse user query
• Extract item code, quantity, warehouse
• Load TDS requirements

Phase 2: Batch Selection

• Query available batches from Bin doctype
• Apply FEFO sorting (First Expired, First Out)
• Select optimal batches for quantity requirement

Phase 3: TDS Compliance

• Validate selected batches against TDS specifications
• Check COA parameters (supports both COA AMB and COA AMB2)
• Report compliance status

Phase 4: Cost Calculation

• Calculate raw material costs
• Include overhead costs
• Provide cost per unit analysis

Phase 5: Optimization (Conditional)

• Only runs if compliance fails
• Suggests alternative batches
• Optimizes blend ratios

Phase 6: Report Generation

• Generate comprehensive workflow report
• Format for Raven channel
• Provide actionable recommendations

Usage Examples

Natural Language Query

"Formulate 1000kg of ITEM_0617027231"

API Usage

from raven_ai_agent.skills.formulation_orchestrator import FormulationOrchestratorSkill

skill = FormulationOrchestratorSkill()
result = skill.handle(
    query="Select batches for product 0617 in warehouse",
    context={
        "quantity_required": 1000,
        "warehouse": "FG to Sell Warehouse - AMB-W",
        "tds_requirements": {
            "pH": {"min": 3.5, "max": 4.5},
            "Brix": {"min": 8.0, "max": 12.0}
        }
    }
)

Inter-Agent Communication

The orchestrator uses the AgentChannel class for communication:

from raven_ai_agent.skills.formulation_orchestrator.messages import AgentChannel

channel = AgentChannel(source_agent="orchestrator")
response = channel.send_to_agent(
    target="batch_selector",
    action="select_batches",
    payload={"item_code": "ITEM_0617027231"}
)

Integration with Raven Channels

The orchestrator can send updates to Raven channels:

from raven_ai_agent.channels import RavenOrchestrator

raven = RavenOrchestrator("formulation-orchestration")
raven.send_workflow_update(
    workflow_id="wf_123",
    current_phase="batch_selection",
    status="completed"
)

Configuration

Environment Variables

• FORMULATION_DEFAULT_WAREHOUSE: Default warehouse for batch queries
• FORMULATION_OVERHEAD_RATE: Overhead rate for cost calculations

ERPNext Settings

• AI Agent Settings doctype for system-wide configuration
• Customer TDS doctype for customer-specific specifications

Version History

Related Skills

• formulation_reader: Data access layer for formulation data
• formulation_advisor: Advisory functions for formulation recommendations