Experiment 2 — brazil-bench (24 cells) · language=rust, model=sonnet, tooling=beads

Run	Rep	Status	code_quality	defect_rate	idiomatic	maintainability	test_coverage	token_efficiency	cost ($)	duration (s)	tokens	turns	Files	LOC
18	rep 1	completed	0.833	0.874	0.770	0.183	0.000	0.056	$1.1087	532.5s	491,969	17	5	1429

src/data.rs (467 loc)

fn normalize_team, fn regex_remove_state_space, fn regex_remove_state_dash, fn team_matches_query, fn normalize_date, fn parse_goals, fn parse_season, fn col_index, fn get_field, fn load, struct Match, struct Player, struct DataStore

src/lib.rs (2 loc)

src/main.rs (227 loc)

fn make_error, fn make_result, fn make_tool_error, fn make_tool_result, fn tools_list, fn handle_request, fn main

src/tools.rs (640 loc)

fn new, fn search_matches, fn get_team_stats, fn head_to_head, fn search_players, fn get_standings, fn get_biggest_wins, fn competition_stats, struct Tools

tests/integration_test.rs (93 loc, test)

fn get_data_dir, fn load_store, fn test_load_datastore, fn test_search_matches_flamengo, fn test_head_to_head_flamengo_fluminense, fn test_search_players_brazil, fn test_get_standings_2019, fn test_get_team_stats_palmeiras_2023, fn test_normalize_team_palmeiras_sp, fn test_normalize_team_flamengo_rj, fn test_normalize_date

language=rust, model=sonnet, tooling=beads

Runs

Code review