Cost Monitoring

---

title: "Cost Monitoring" tier: "3" status: "verified" version: "1.1.165" last-verified: "2026-06-11" commit: "b1c347b1" network-badge: "none" risk-tags:

• "RISK: MEDIUM"

---

💰 Cost Monitoring

This page documents the cost-accounting dashboard for monitoring API expenses, token consumption, and query latency across all LLM providers.

---

1. Neural Footprint & Token Cost Monitoring (Command_Table)

The Command_Table component is embedded in the Oversight page and provides a detailed breakdown of every LLM interaction.

1.1 Interactive Elements

• [Table Row] Interaction Logs

  • Default state: Populated with recent prompt-response pairs.
  • Visible location: Center region of the Command_Table panel.
  • Observable side effect: Clicking a row expands the panel to show raw request and response body text.
  • Keyboard shortcut: None.
  • Missing in build: Always present.

• [Column Header] Latency Tracker

  • Default state: 0ms (or latency of the most recent query).
  • Visible location: Upper-right column header of Command_Table.
  • Observable side effect: Displays execution durations in milliseconds to pinpoint slower local or cloud models.
  • Keyboard shortcut: None.
  • Missing in build: Always present.

• [Cell] Token Efficiency

  • Default state: 0 tokens (accumulated).
  • Visible location: Inside each query row of Command_Table.
  • Observable side effect: Details prompt and response token lengths to help optimize prompt templates.
  • Keyboard shortcut: None.
  • Missing in build: Always present.

---

2. Budget Gauges & Cost Dashboard

2.1 Interactive Elements

• [Gauge] Budget Quotas

  • Default state: 0.0 USD used.
  • Visible location: Upper metrics grid of the Oversight dashboard.
  • Observable side effect: Dynamically reflects real-time cumulative LLM API expenses against configured caps.
  • Keyboard shortcut: None.
  • Missing in build: Always present.

• [Gauge] Total Spend

  • Default state: 0.00 USD.
  • Visible location: Oversight dashboard or Cost analytics panel.
  • Observable side effect: Summarizes global API expenses across all agents and models.
  • Keyboard shortcut: None.
  • Missing in build: Always present.

• [Metric] Daily Burn Rate

  • Default state: 0.00 USD/day.
  • Visible location: Upper row of the Cost dashboard metrics.
  • Observable side effect: Projects expected daily spending based on recent model requests.
  • Keyboard shortcut: None.
  • Missing in build: Always present.

• [Table] Provider Cost Breakdown

  • Default state: Lists configured providers and their aggregated costs.
  • Visible location: Lower panel of the Cost dashboard.
  • Observable side effect: Allows operators to identify which LLM services are consuming the most tokens/credits.
  • Keyboard shortcut: None.
  • Missing in build: Always present.

• [Button] Export Cost Report

  • Default state: Enabled.
  • Visible location: Top right of the Cost dashboard panel.
  • Observable side effect: Triggers a download of cumulative cost telemetry as a CSV spreadsheet.
  • Keyboard shortcut: None.
  • Missing in build: Yes (planned control, missing in current build).

• [Selector] Date Range

  • Default state: Last 30 Days.
  • Visible location: Top toolbar of the Cost dashboard.
  • Observable side effect: Updates all gauges, graphs, and tables to reflect cost telemetry in the chosen time window.
  • Keyboard shortcut: None.
  • Missing in build: Always present.

2.2 Per-Agent Budget Tracking

Each agent tracks its own cost accumulation via RunContext::current_cost_usd. When the cumulative cost exceeds RunContext::budget_usd, the runner raises BudgetExhausted and the mission fails gracefully.

---

3. Cost Optimization Tips for SMBs

Strategy	Impact	How To
Use local models (Ollama)	$0.00 per query	Install Ollama + llama3:8b, set OLLAMA_HOST
Lower reasoning depth	Fewer tokens per turn	Set Mythos Reasoning Depth to 1-4 for simple tasks
Set tight budgets	Prevents runaway spend	DEFAULT_AGENT_BUDGET_USD=0.50
Use auto-approve for reads	Fewer manual approvals	Keep AUTO_APPROVE_SAFE_SKILLS=true
Review token efficiency	Find wasteful prompts	Inspect Command_Table token counts per query

Tip

For a typical SMB running 10 tasks/day with a mix of local and cloud models, expect monthly cloud API costs of $5–$50 USD depending on model selection and task complexity.

---

4. Telemetry & Accumulation

The engine tracks tokens globally via GovernanceHub::tpm_accumulator (gov.rs:L38) (Tokens Per Minute), which feeds into the dashboard metrics:

Metric	Source	Description
TPM (Tokens Per Minute)	GovernanceHub::tpm_accumulator (gov.rs:L38)	Global token throughput counter
Per-run cost	RunContext::current_cost_usd	Cumulative cost for the active agent run
Budget cap	RunContext::budget_usd	Maximum spend allowed for the run

---

5. Local-First Pledge

Important

100% Local-First / Zero Data Leaks.

• All token counting, spending accumulation, and budget monitoring processes occur 100% locally.
• No usage statistics or billing telemetry is ever transmitted to external trackers.

---

→ See Approvals-&-Quotas for budget enforcement and failure modes. → See Configuration for engine limit settings (DEFAULT_AGENT_BUDGET_USD, MAX_TASK_LENGTH). → See LLM-Providers for provider failover thresholds.

---

Complete Lexicon: For the authoritative technical breakdown, see the main repository GLOSSARY.md. Every [[Glossary#term|term]] link on this page resolves to an entry there.