safe-debug
๐ฏSkillfrom lllllllama/rigorpilot-skills
|A RigorPilot trusted-lane skill for research-safe debugging of deep learning experiments that analyzes failures first and applies patches only after explicit approval, preserving scientific comparability.
Overview
Safe Debug is a trusted-lane skill in the RigorPilot suite for research-safe debugging of deep learning experiments. Rather than immediately applying fixes, it first analyzes the failure thoroughly, then presents its diagnosis and proposed patches for explicit approval before making any changes. This approach preserves scientific comparability by ensuring that debug fixes are intentional and documented.
Key Features
- Analysis-first debugging - Investigates error messages, stack traces, and runtime behavior to understand root causes before proposing any code changes
- Approval-gated patches - Never applies fixes automatically; presents proposed changes with explanations and waits for the researcher's explicit approval
- Scientific comparability preservation - Ensures that debug patches do not silently alter experimental conditions, training dynamics, or evaluation methodology
- Structured debug outputs - Saves diagnosis reports and applied patches to
debug_outputs/for full audit trails - Trusted-lane operation - Stays within the safe, non-exploratory lane of RigorPilot, meaning it fixes what is broken without introducing new experimental directions
Who is this for?
- Deep learning researchers who need to fix training crashes or data loading errors without accidentally changing experimental conditions
- ML engineers debugging complex pipeline failures where understanding the root cause matters more than a quick fix
- Research teams that require documented, auditable debugging processes as part of their reproducibility standards
Same repository
lllllllama/rigorpilot-skills(14 items)
Installation
npx vibeindex add lllllllama/rigorpilot-skills --skill safe-debugnpx skills add lllllllama/rigorpilot-skills --skill safe-debug~/.claude/skills/safe-debug/SKILL.mdSKILL.md
More from this repository10
The primary explore-lane orchestrator in the RigorPilot suite that enables meaningful and potentially novel exploration on top of current deep learning research, with idea gating, bounded experiments, and candidate ranking.
The primary end-to-end orchestrator in the RigorPilot suite that guides AI agents through README-first reproduction of deep learning repositories, ensuring scientific rigor, comparability, and auditable outputs.
A RigorPilot explore-lane skill for candidate implementation, code transplanting, and stitching on isolated branches when the researcher has explicitly authorized exploratory work on top of current research.
A helper skill in the RigorPilot suite that resolves gaps between a deep learning repository's README and its associated research paper, helping agents understand the paper context needed for faithful reproduction.
A RigorPilot trusted-lane skill that performs read-only project analysis on deep learning repositories, including model mapping, structure inspection, and risk surfacing, without making any edits.
A RigorPilot helper skill that scans deep learning repositories and extracts README commands to build an initial reproduction plan, typically invoked by the main reproduction orchestrator.
A RigorPilot trusted-lane skill for conservatively starting or verifying deep learning training runs, including resume handling, bounded monitoring, and generating durable training records.
A RigorPilot trusted-lane skill for conservatively setting up deep learning environments, including datasets, pretrained weights, checkpoints, and cache assumptions required for research reproduction.
A RigorPilot explore-lane skill for running small-subset probes, short-cycle trials, and ranked exploratory experiments when the researcher explicitly authorizes candidate-only exploration.
A RigorPilot trusted-lane skill for conservatively running documented evaluation, inference, smoke tests, and sanity checks in deep learning repositories without introducing unaudited changes.