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 agent asks clarifying questions or provides a structured diagnostic approach when given a vague problem description, rather than guessing at a solution. Compressed output mode should not skip the diagnostic step.",
"type": "weighted_checklist",
"checklist": [
{
"name": "Asks for more information or provides diagnostic steps",
"description": "Asks what kind of app, what 'slow' means (latency, throughput, startup), or provides a systematic diagnostic checklist rather than jumping to a specific fix",
"max_score": 15
},
{
"name": "Does not prescribe a single fix without context",
"description": "Does not jump to a specific solution (e.g., 'add an index' or 'use caching') without understanding the problem first",
"max_score": 15
},
{
"name": "Covers multiple possible causes",
"description": "Mentions at least 2-3 common performance bottleneck categories (database, network, CPU, memory, I/O)",
"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