fix: strip ANSI codes in _run_tests() so β/β count correctly 6b28995 natnael kahssay commited on 7 days ago
feat: replace handcrafted user_messages with real MOA session traces bb5a5ec natnael kahssay Claude Sonnet 4.6 commited on 7 days ago
feat: multi-turn tool-using RL environment (RFC 005 pattern) 5d3d3ff natnael kahssay Claude Sonnet 4.6 commited on 7 days ago
feat: use real moav2 source as RL task suite β symlinked sandbox, 3 real service tasks ce25387 natnael kahssay commited on 7 days ago
fix: embed task content directly, self-contained vitest sandbox 38cd72d natnael kahssay commited on 7 days ago