💰 Nur PROMETHEUS
Financial Intelligence Agent | Domain: Finance
Bearer of light for financial operations, cost analysis, budgeting, and economic insights.
Overview​
KOSMOS V2.0 Nur PROMETHEUS Agent - Financial Intelligence
Nur PROMETHEUS handles all financial operations including budgeting, cost tracking, financial analysis, and cost governance.
Upgraded to LangGraph base with:
- State persistence (checkpointing)
- Pentarchy voting integration (cost-benefit analysis)
- Semantic router integration
- SDUI financial visualization components
- Real-time cost tracking
This documentation is automatically extracted from source code.
Source: implementation/backend/agents/nur_prometheus.py (903 lines)
Enumerations​
CostCategory​
Cost categories for tracking.
| Member | Value |
|---|
BudgetPeriod​
Budget period types.
| Member | Value |
|---|
ApprovalDecision​
Cost approval decisions.
| Member | Value |
|---|
FinanceOperation​
Financial operations.
| Member | Value |
|---|
AlertLevel​
Budget alert levels.
| Member | Value |
|---|
NurPrometheusState​
State for Nur PROMETHEUS financial workflow.
Extends base state with finance-specific fields.
Inherits from: AgentGraphState
NurPrometheusAgent​
Nur PROMETHEUS - Financial Intelligence Agent
Responsibilities:
- Estimate costs before execution
- Track actual costs and usage
- Enforce budget limits
- Generate financial reports
- Cost-based governance decisions
- Participate in Pentarchy voting (ROI analysis)
- Alert on budget thresholds
MCP Servers:
- postgres-mcp: Cost tracking database
SDUI Components:
- GlassMeter: Budget utilization
- GlassChart: Cost trends
- GlassCard: Financial summary
Inherits from: <ast.Subscript object at 0x7fe8021b0670>
Methods​
__init__()​
create_state_class() → type​
Return NurPrometheusState for workflow.
define_nodes() → Dict[str, Callable]​
Define Nur PROMETHEUS-specific workflow nodes.
define_edges() → List[tuple]​
Define Nur PROMETHEUS-specific workflow edges.
async vote_on_proposal(proposal: Dict[str, Any]) → Dict[str, Any]​
Cast vote in Pentarchy governance.
Nur PROMETHEUS votes based on cost-benefit analysis.
estimate_llm_cost(model: str, input_tokens: int, output_tokens: int) → Decimal​
Estimate cost for LLM inference.
estimate_mcp_cost(server: str, call_count: int) → float​
Estimate cost for MCP tool calls.
async quick_estimate(model: str, input_tokens: int, output_tokens: int) → Dict[str, Any]​
Quick cost estimate without full workflow.
async check_can_proceed(estimated_cost: float) → Dict[str, Any]​
Quick check if an operation can proceed given cost.