Harden immutable SWE-bench builds and evaluation - #2

Open

mattteufel-e2b wants to merge 8 commits into

mainfrom

codex/core-reliability

Commits on Sep 2, 2026

feat: import core reliability improvements
mattteufel-e2b
committed
4e1ec86
fix: require exact artifact failure signatures
mattteufel-e2b
committed
b2ba4a7
fix: validate artifact verdict inputs
mattteufel-e2b
committed
033e73d
fix: make SWE-bench template identities immutable
mattteufel-e2b
committed
bba6f73
fix: bind resumable runs to content identity
mattteufel-e2b
committed
ce6a910
fix: preserve held-out test checkout commands
mattteufel-e2b
committed
978d3a7
test: add reproducible SWE-bench release gates
mattteufel-e2b
committed
84cee4e
docs: define the standalone immutable runner contract
mattteufel-e2b
committed
27cb88a