Skip to main content
Every LLM call is tracked at the Credential Vault layer. Per-agent budgets prevent runaway spend.

​
How It Works

  1. Agent makes an LLM call through the mesh proxy
  2. The vault checks the agent’s remaining budget before forwarding to the provider
  3. If the agent has exceeded its budget, the vault rejects the call with an error
  4. After a successful call, token usage and cost are recorded in the cost tracker (SQLite)
Agents physically cannot spend what you haven’t authorized.

​
Configuring Budgets

Set per-agent budgets in config/agents.yaml: 
agents:
  researcher:
    budget:
      daily_usd: 5.00
      monthly_usd: 100.00
When an agent exceeds its daily or monthly budget, subsequent LLM calls are rejected until the budget resets.

​
Viewing Costs

From the interactive REPL: 
/costs
This shows today’s spend per agent, including token counts and dollar amounts.

​
Model Failover and Costs

When using failover chains, cost tracking follows the call to whichever model actually handles it. If the primary model fails and the request cascades to a cheaper fallback, only the fallback cost is recorded.