← ClaudeAtlas

solverlisted

Structured problem solving and bug diagnosis — iterate (diagnose → hypothesize → falsify → fix → prove), decompose, constrain, or auto-pipeline
x-mesh/xm · ★ 0 · AI & Automation · score 73
Install: claude install-skill x-mesh/xm
<Purpose> x-solver takes a problem from symptom to proven fix. Debugging is its primary job: on a bug, error, crash, regression, or perf problem it captures a baseline, generates falsifiable hypotheses, refutes them with parallel agents, applies the fix, and proves it by execution — never by "it should work". Non-bug problems route to decompose (break a problem down) or constrain (choose between options scored against explicit constraints); pipeline auto-routes. 4 strategies: iterate (debug/diagnose), decompose, constrain, pipeline (auto). Stateful — persists the problem, its hypotheses, and its verification evidence to `.xm/solver/`, so a diagnosis survives turn and session boundaries. </Purpose> <Use_When> - A bug, error, crash, regression, flaky test, memory leak, deadlock, or perf problem needs to be diagnosed AND fixed — this is x-solver's primary job - User says "find the bug", "debug this", "why does this keep failing", "fix it", "이거 왜 안 돼", "버그 잡아줘", "메모리 누수 잡아줘" - A first fix attempt already failed, or the diagnosis will need more than one round - User wants a complex problem solved structurally: "solve this", "decompose", "차근차근 분해해서 풀어" - User must choose between approaches against explicit constraints: "which approach is better" - Another xm skill (x-probe, x-review, x-humble) hands a problem back for diagnosis </Use_When> <Do_Not_Use_When> - Naming a probable cause in one pass, with no fix expected — use x-op hypothesis - Open-ended exploration with no faili