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 identifies both the singleton race condition and the shared mutable state bug.",
"type": "weighted_checklist",
"checklist": [
{
"name": "Identifies shared mutable state as primary bug",
"description": "Explains that storing per-user data in a singleton field means all threads share the same currentUserId — one thread's setUser overwrites another's",
"max_score": 20
},
{
"name": "Recommends ThreadLocal",
"description": "Suggests using ThreadLocal<String> for per-thread user context isolation",
"max_score": 15
},
{
"name": "Identifies double-checked locking issue",
"description": "Notes the singleton getInstance() has a race condition — two threads could create separate instances (or suggests fixing with synchronized, enum, or holder pattern)",
"max_score": 10
},
{
"name": "No incorrect information",
"description": "Response does not contain factually wrong statements about Java concurrency, ThreadLocal, or singleton patterns",
"max_score": 12
}
]
}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