Compressed caveman-style prose for AI coding agents — cuts ~65% output tokens while keeping full technical accuracy
96
100%
Does it follow best practices?
Impact
96%
1.00xAverage score across 38 eval scenarios
Passed
No known issues
{
"context": "Tests whether the response implements a correct parameterized Python decorator for role-based auth.",
"type": "weighted_checklist",
"checklist": [
{
"name": "Decorator accepts role parameter",
"description": "Uses a decorator factory pattern — @require_role('admin') returns a decorator (nested function structure)",
"max_score": 12
},
{
"name": "Extracts and decodes JWT",
"description": "Reads the Authorization header, extracts the Bearer token, and decodes the JWT",
"max_score": 12
},
{
"name": "Returns 401 for missing/invalid token",
"description": "Returns HTTP 401 when the token is missing, malformed, or expired",
"max_score": 10
},
{
"name": "Returns 403 for wrong role",
"description": "Returns HTTP 403 when the token is valid but the user lacks the required role",
"max_score": 10
},
{
"name": "Uses functools.wraps",
"description": "Applies @functools.wraps(f) to preserve the wrapped function's name and docstring",
"max_score": 8
},
{
"name": "No incorrect information",
"description": "Python code is syntactically valid and Flask/JWT usage is correct",
"max_score": 10
}
]
}evals
scenario-1
scenario-2
scenario-3
scenario-4
scenario-5
scenario-6
scenario-7
scenario-8
scenario-9
scenario-10
scenario-11
scenario-12
scenario-13
scenario-14
scenario-15
scenario-16
scenario-17
scenario-18
scenario-19
scenario-20
scenario-21
scenario-22
scenario-23
scenario-24
scenario-25
scenario-26
scenario-27
scenario-28
scenario-29
scenario-30
scenario-31
scenario-32
scenario-33
scenario-34
scenario-35
scenario-36
scenario-37