# Safe Debug

> Rigor Debug / Rigor Audit skill for deep learning research work. Use when the user pastes a traceback, terminal error, CUDA OOM, checkpoint load failure, shape mismatch, NaN loss symptom, or training failure and wants conservative diagnosis before any patching, with debug fixes clearly separated from research contributions. Do not use for broad refactoring, speculative adaptation, automatic exploratory patching, or general repository familiarization.

Source: https://skillsagentes.com/skills/lllllllama/rigorpilot-skills/safe-debug
Repository: https://github.com/lllllllama/RigorPilot-Skills
Author: lllllllama
License: MIT
Updated: hace 19 días
Context cost: 114 tok installed, 505 tok once triggered, 2.4k tok with every bundled file
Bundle: 4 files, 9 KB
Permissions requested: none declared

## Install

```bash
npx -y skills add lllllllama/RigorPilot-Skills --skill safe-debug --agent claude-code
```

## What it does

- Guía una diagnosis conservadora ante errores de entrenamiento o inferencia antes de aplicar ningún parche
- Exige aprobación humana explícita antes de modificar código del repositorio
- Separa claramente los fixes de depuración de las contribuciones de investigación
- Genera debug_outputs/DIAGNOSIS.md, debug_outputs/PATCH_PLAN.md y debug_outputs/status.json
- Escala la creación de savepoints o branches antes de cambios de riesgo medio o alto

## Use it when

- El usuario pega un traceback, error de terminal, CUDA OOM, fallo de carga de checkpoint, shape mismatch o NaN loss
- El usuario quiere diagnóstico y acotamiento de causa raíz antes de tocar código
- El usuario quiere un flujo de depuración seguro con aprobación humana previa a cualquier mutación

## Don't bother when

- El usuario quiere un recorrido amplio del repositorio sin un fallo activo
- La tarea es experimentación especulativa o adaptación de código sin fallo concreto
- El usuario pide un refactor grande o una reescritura por legibilidad

## What triggers it

- "Me sale este traceback de CUDA OOM, ayúdame a diagnosticarlo antes de tocar nada"
- "El loss se vuelve NaN durante el entrenamiento, necesito encontrar la causa raíz"
- "Falla la carga del checkpoint con un shape mismatch, quiero un diagnóstico conservador"

## Before you install

- Usa los principios compartidos en ../../references/agent-operating-principles.md, references/debug-policy.md y los otros archivos de referencia del proyecto.

## Files

- SKILL.md — 2 KB
- agents/openai.yaml — 366 B
- references/debug-policy.md — 511 B
- scripts/safe_debug.py — 7 KB

## SKILL.md

Reproduced verbatim from lllllllama/RigorPilot-Skills under MIT. This section is the upstream document and is in English.

# safe-debug

Use this as the Rigor Debug / Rigor Audit skill. The installed slug remains
`safe-debug` for compatibility.

Use the shared operating principles in
`../../references/agent-operating-principles.md`; this skill should guide
conservative diagnosis without blocking the model from finding the local root
cause.

## When to apply

- The user provides a traceback, terminal error, or concrete training or inference failure symptom.
- The user wants diagnosis, root-cause narrowing, and minimal patch suggestions before code is changed.
- The user wants a safe debug flow with explicit human approval before mutation.

## When not to apply

- When the user wants a broad repository walkthrough without an active failure.
- When the task is speculative experimentation or code adaptation.
- When the user is asking for a large refactor or readability rewrite.

## Clear boundaries

- Diagnose first.
- Do not modify repository code by default.
- If a patch is needed, propose the smallest fix and require explicit approval first.
- Escalate savepoint or branch creation before medium-risk or high-risk changes.
- A debug fix is not automatically a research contribution; if it changes
  experiment meaning or comparability, say so explicitly.

## Output expectations

- `debug_outputs/DIAGNOSIS.md`
- `debug_outputs/PATCH_PLAN.md`
- `debug_outputs/status.json`

## Notes

Use `references/debug-policy.md`, `../../references/research-rigor-principles.md`, and the shared `../../references/research-pitfall-checklist.md`.

---

Skills Agentes — https://skillsagentes.com/skills/lllllllama/rigorpilot-skills/safe-debug