Scored across 12 tools
Tools mostly target distinct resources, and the descriptions explicitly cross-reference siblings (get_next_item for the one packet, list_worklist for the ranking, get_function_brief once chosen; list_coupled_files vs list_duplicate_functions). There is mild clustering around 'what to fix' (list_worklist, get_next_item, get_function_brief, list_claims) and around cross-run analysis (get_trend, get_function_history, get_ratchet_report), but each boundary is spelled out.
Every tool follows a clean verb_noun snake_case pattern: list_* (list_runs, list_worklist, list_coupled_files, list_duplicate_functions, list_claims), get_* (get_next_item, get_trend, get_function_brief, get_function_history, get_ratchet_report), and check_* (check_config, check_gate). No mixed conventions or vague verbs.
12 tools sits squarely in the well-scoped 3-15 range and each maps to a real operation in the crapkit workflow (discover, survey, pick, inspect, gate-check, audit debt). No obvious filler or redundant entries.
The surface covers the analysis lifecycle: runs, config validation, work selection, per-function detail, trends, coupling, duplication, ratchet debt, gate checks, and claims listing. Gaps exist — no tool writes a claim (explicitly CLI-only) and no tool marks the baseline run — but these are documented and workable around.
