syntaxixr/receipts-study
Viewer • Updated • 984 • 78 • 1
Did the coding agent really finish
Is the agent's work really done? Tests that never ran on old code (Receipts) and /goal saying "met" while hidden checks fail (goalpost).
Note 181 real changes: would the tests have failed without the fix?
syntaxixr/goalpost-benchmark
Viewer • Updated • 33 • 38Note 33 runs: the built-in /goal judge said met 17 of 17 times, 8 were broken