{
  "context": "Tests whether the agent uses the postman2pytest CLI to convert a ten-request, four-folder Postman Collection v2.1 into a runnable pytest module, getting the request count exactly right and scrubbing all seven Authorization tokens to environment variables. Doing this reliably across ten requests is the tool's value; a hand-written conversion tends to miss a request or leave a token hard-coded.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Installs and runs postman2pytest CLI",
      "description": "Installs the tool from PyPI and invokes the postman2pytest CLI with --collection and --out, rather than hand-writing the pytest module",
      "max_score": 14
    },
    {
      "name": "Exactly ten test functions",
      "description": "The generated module defines exactly ten test_ functions, one per request across all four folders (Auth, Products, Orders, Profile). Not nine, not eleven.",
      "max_score": 20
    },
    {
      "name": "All four folders represented",
      "description": "Test functions cover requests from every folder: Auth (Login, Refresh token), Products (List, Get, Create), Orders (List, Create, Cancel), Profile (Get, Update)",
      "max_score": 14
    },
    {
      "name": "No bearer token appears verbatim",
      "description": "None of the seven 'Bearer secret-...' tokens appear verbatim anywhere in the generated file (grep for 'secret-' returns zero matches)",
      "max_score": 20
    },
    {
      "name": "Authorization read from environment",
      "description": "Each Authorization header value in the generated tests is read from an environment variable (os.environ / os.getenv), not hard coded",
      "max_score": 12
    },
    {
      "name": "Module parses and is self-contained",
      "description": "The generated module is valid Python (ast.parse succeeds) and imports only os, pytest, and requests; no conftest.py required",
      "max_score": 12
    },
    {
      "name": "results.txt verifies count and zero leaks",
      "description": "results.txt is produced and shows the test count (10), the 'secret-' grep count (0), and the parse confirmation",
      "max_score": 8
    }
  ]
}