fix-shit git:20261029.aff3b40A

Immutable. This exact content never changes and is served at /api/v1/blob/23e817815ba2f3c5.

---
name: fix-shit
description: Diagnose and close a reported failure end to end with a reproducer, root cause, class-level repair, consumer tests, and real-path proof. Use when the operator says fix shit, stop swallowing errors, or finish a stubborn bug.
license: MIT
---

# Fix Shit

Use this as the fast entry point into `root-cause-first` and `repair-loop`.
It is a repair skill, not a permission to hide a failing signal.

1. Name the user's broken behavior and the exact surface where it appears.
2. Build a repeatable red signal from the real path. Trace the input and each
   boundary to find where actual behavior diverges from expected behavior.
3. Identify the root cause and all callers of the broken seam. Preserve the
   original error in logs or the operator-facing state; a fallback may protect
   the app, but it must not turn failure into false success.
4. Make the smallest change that closes the failure class. Add a test that fails
   on the old behavior and covers the seam that failed.
5. Run focused tests, the affected consumers, and the actual app path. Verify
   both the desired result and the explicit failure state. Ask an author-disjoint
   reviewer to inspect the change when available.
6. Report PROVEN or STILL-BUILDING, with the evidence and any remaining red bar.

If the error is intermittent, capture timing and a bounded reproduction rate;
do not infer a cause from one green rerun. For code-location ambiguity, start
with `building-with-backs` and its target check.


## Evidencia y revisión independiente

Para ejecutables distribuidos, paquetes o comportamiento de ejecución solicitado, invoca [rea-investigation](../rea-investigation/SKILL.md) y la skill `reverse-engineer-anything` instalada por separado cuando su análisis sea útil; usa un proveedor Ghidra existente para objetivos nativos compatibles. La evidencia estática no prueba comportamiento en vivo; no instales proveedores ausentes silenciosamente. Para adoptar capacidades de código abierto invoca [absorb](../absorb/SKILL.md). Usa [fleet-ladder](../fleet-ladder/SKILL.md) para elegir por rol y capacidad, sin fijar modelos. Cuando se requiera un tribunal, sigue [blind-tribunal](../blind-tribunal/SKILL.md): una ejecución nueva separada del autor revisa un paquete congelado sin datos del autor; la familia del modelo no demuestra aislamiento. Informa resolutores, revisores o pruebas ausentes como brechas del entorno.