Re:Show HN: Loki Mode hit 99.67% SWE-Bench – MAF built a Saa...
by slogansand on Wednesday January 07, 2026 @11:16AM
(#46528163)
Author here. Quick context on the benchmarks:
We used RARV (Retrieve, Analyze, Reason, Validate) pattern with multi-agent collaboration. Each problem gets worked by specialized agents, reviewed by 3 parallel reviewers (code, business logic, security), and only merged after consensus.
The 99.67% isn't cherry-picked. Full run against standard SWE-Bench dataset. Happy to share methodology if anyone wants to reproduce.
We used RARV (Retrieve, Analyze, Reason, Validate) pattern with multi-agent collaboration. Each problem gets worked by specialized agents, reviewed by 3 parallel reviewers (code, business logic, security), and only merged after consensus.
The 99.67% isn't cherry-picked. Full run against standard SWE-Bench dataset. Happy to share methodology if anyone wants to reproduce.