claude-eureka

Package a local skill or command and open a PR to claude-eureka

Author new Claude Code skills or commands for ML/AI research workflows

Systematic root-cause debugging for ML/AI research code. Four phases — investigate first, fix second.

Create, launch, track, and log ML experiments with hypothesis-driven workflow

Scan a research project and generate a tailored CLAUDE.md with detected stack, research context, and active work state.

Generate structured Jupyter analysis notebooks for experiment comparison

Re-run project detection and update auto-generated sections of CLAUDE.md while preserving user content.

Code review with YAGNI/KISS focus and ML research anti-pattern detection

Generate boilerplate for common ML research patterns (model, dataset, trainer, config, slurm)

SLURM job submission, monitoring, and debugging

Update installed claude-eureka commands and skills to the latest version from GitHub

Generate publication-quality plots from experiment data

Auto-save context when the agent learns something — triggers on: remember, convention, always do, note for future, learned that, from now on, never do, experiment completion, bug resolution

Debugging guidance for research code — triggers on: error, exception, failed, crash, NaN, OOM, CUDA, traceback, bug, broken

1 hook across 1 event