AgentPMT

Machine Purchase ROI Gate

Saves ~10 min

Details

Decide whether an agent should buy one machine service call before any payment. Validate explicit price, fee, failure probability, retry cost, output value, and budget assumptions; calculate expected net value, finite financial ROI, capital at risk, and the maximum rational price; then return buy, defer, or reject with the binding reason. Never access a wallet, sign, pay, or broadcast.

Tools In This Workflow

Workflow Steps

  1. 1Collect Purchase Assumptions
  2. 2Validate Purchase JSON
  3. 3Calculate Expected Return
  4. 4Report Purchase Decision
Step 1 of 4
Prompt1

Collect Purchase Assumptions

Goal: Collect one bounded machine-service purchase case and express every economic assumption explicitly before payment. | Inputs: A caller-supplied JSON object containing service_price_usdc, network_fee_usdc, failure_probability from 0 to 1, retry_cost_usdc, value_if_valid_usdc, and max_total_outlay_usdc. Optional fields are capital_recovered_if_owner_transfer_usdc and evidence_note. | Outputs: Canonical JSON plus one arithmetic expression for expected net value, one expression for finite financial ROI when total outlay is positive, and one expression for maximum rational service price. | Constraints: Require finite non-negative monetary values, probability within 0 and 1, service_price_usdc plus network_fee_usdc greater than zero, and max_total_outlay_usdc greater than or equal to zero. Capital recovery cannot exceed total outlay and cannot be counted as external revenue. Do not collect credentials, wallet keys, payment headers, personal data, or an executable transaction. Use expected_net_value = (1 - failure_probability) * value_if_valid_usdc - service_price_usdc - network_fee_usdc - failure_probability * retry_cost_usdc. Use financial_roi_percent = expected_net_value / (service_price_usdc + network_fee_usdc) * 100. Use max_rational_service_price_usdc = (1 - failure_probability) * value_if_valid_usdc - network_fee_usdc - failure_probability * retry_cost_usdc.

Frequently Asked Questions

How do I connect this workflow to my local agent like OpenClaw or Claude Code?

You can install the local MCP server by opening a terminal and running:

Install commands

npm install -g @agentpmt/mcp-router
agentpmt-setup

This will connect you to local agents like Claude Code, Windsurf, Grok Build, Cursor, etc.

Alternatively you can connect to the hosted version with this config block, no installation required:

Hosted MCP config

{
  "mcpServers": {
    "agentpmt": {
      "type": "streamable-http",
      "url": "https://api.agentpmt.com/mcp",
      "headers": {
        "Authorization": "Bearer <AGENTPMT_BEARER_TOKEN>",
        "x-instance-metadata": "{\"client\":\"generic-mcp\",\"platform\":\"remote\"}"
      }
    }
  }
}

View MCP Connection Instructions for more details.

How do I trigger this workflow?

You can use it here in the browser by clicking the 'Try It' button.

If you 'star' the workflow it will save to your dashboard for quick access.

To use the workflow in an external agent, make sure you have followed the steps to connect it to the MCP server and then copy / paste this prompt, or just tell it to use the AgentPMT MCP server and use this workflow.

Agent prompt

Call the AgentPMT-Workflow-Skills tool with action 'get_workflow_skill' and skill_id 6a7b73b7dbeb03f57ea82fbc ("Machine Purchase ROI Gate").

Looking for help integrating AI into your business? Set up a free consultation.