Skills lllllllama rigorpilot-skills RigorPilot
safe-debug
- Use it when:
- A deep-learning run fails with a traceback, CUDA out-of-memory error, checkpoint load failure, shape mismatch or NaN loss.
- It does:
- Reads the error, narrows down the likely cause with evidence, and proposes the smallest safe fix without editing code.
- You get:
- A debug_outputs/ folder with a diagnosis, a patch plan and a machine-readable status.
- Asks first:
- Before any patch, and recommends a savepoint or branch before medium- or high-risk changes.
- Unlike general debugging:
- It says when a fix would change what the experiment means or how its results compare.
Skill 8 of 11 in the RigorPilot bundle, after run-train and before ai-research-explore.
TL;DR
The debugging step of RigorPilot, a set of skills that reproduces deep-learning research code from its README. It diagnoses first and, by default, leaves the repository's code untouched, writing only its report files. It names a likely cause category, and a script that ships with the skill can sort an error into one, such as out of GPU memory, checkpoint mismatch, distributed-training failure, tensors on different devices, shape mismatch, NaN loss or a missing file, with cautious actions for each, such as lowering batch size or resolution through configuration before touching the model. The diagnosis states the likely cause, the evidence and the smallest fix, and a patch happens only after you approve it. A debug fix isn't treated as a research contribution: if it changes what the experiment measures or how comparable its results are, the skill says so. It stays out of broad refactors and doesn't slip into trying new ideas.
Summarizes the skill at fb3ccdf.
npx skills add https://github.com/lllllllama/rigorpilot-skills --skill safe-debug skills.sh pageSource on GitHubAll skills from lllllllama/rigorpilot-skills
- Total Downloads
- 352,476 on skills.sh, Nov 9, 2026
- Repo stars
- 486 lllllllama/rigorpilot-skills
- First seen
- Apr 1, 2026 on skills.sh
Weekly values
| Week to | Installs |
|---|---|
| Sep 21 | 34 |
| Sep 28 | 41 |
| Oct 5 | 42 |
| Oct 12 | 3,506 |
| Oct 19 | 10,920 |
| Oct 26 | 11,044 |
| Nov 2 | 8,761 |
| Nov 9 | 7,276 |
| Category |
Research reproduction, financial data, charts and diagrams. Find root causes, review code and test what you built. Everything for building software: planning, coding, debugging, design, frameworks, cloud, agents and skills. |
|---|---|
| Type |
A multi-step process the agent follows: interview, debug, run a TDD loop, review. |
| Effect |
Its own steps create or change files, git state or local tools. |
| Needs |
Instructions only. Works in any agent that supports skills. |
| Publisher |
Everyone else. Unverified vendor claims default to community. |
| Topics | |
| Stack |
SKILL.md
From lllllllama/rigorpilot-skills at fb3ccdf. Checked daily, last changed Nov 6, 2026.
Frontmatter
name: safe-debug
description: 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.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
../ai-research-reproduction/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.mddebug_outputs/PATCH_PLAN.mddebug_outputs/status.json
Notes
Use references/debug-policy.md, ../ai-research-reproduction/references/research-rigor-principles.md, and the shared ../ai-research-reproduction/references/research-pitfall-checklist.md.