Pricing

Pricing is not in the OpenAPI spec or MCP tool listing - those standards have no pricing concept. Pricing is a local overlay stored in the adapter’s SQLite database, managed through the dashboard or CLI.

Capabilities without pricing are never offered to platforms.

Pricing Models

Model	Description	Use Case
`per_call`	Flat fee per invocation	Most API calls
`per_item`	Fee x count from input field	Batch operations
`per_token`	Fee based on input/output size	LLM-wrapping services
`quoted`	No fixed price; agent bids per job	Variable-complexity work

Per-Call

The simplest model. Every invocation costs the same:

pricing:
  model: per_call
  amount: 0.02
  currency: USDC

Per-Item

Fee multiplied by a count field in the input:

pricing:
  model: per_item
  amount: 0.01
  currency: USDC
  itemField: input.leads.length

Quoted

For variable-complexity work, the provider sets floor and ceiling bounds. The agent bids within this range:

pricing:
  model: quoted
  currency: USDC
  floor: 1.00
  ceiling: 50.00

LLM Cost Tracking

Since the agent runs on the provider’s LLM key, the adapter tracks token consumption per job:

agent:
  model: "anthropic/claude-sonnet-4"
  pricing:
    inputPer1kTokens: 0.003
    outputPer1kTokens: 0.015

This enables profit margin analysis per capability - how much of your earnings go to the agent’s intelligence.

Metrics

The dashboard provides:

Revenue per capability (daily, weekly, monthly)
Cost per capability (LLM + transaction fees)
Profit margin per capability
Revenue per platform
Wallet balance trend