agents-toolkit

Use when working with datasets, creating data directories, or building descriptive statistics notebooks. Triggers on data loading, dataset setup, or exploratory data analysis tasks.

Use when creating, editing, or executing Jupyter notebooks (.ipynb files). Covers jupytext pairing, notebook execution, and output verification.

Use when creating plots, charts, or visualizations with matplotlib or seaborn. Defines plotting style standards.

Review a session plan for scope creep and unnecessary complexity before execution begins. Combines scope-check and simplicity-audit. Run after Plan Mode, before any code is written.

Use before any creative work — creating features, building components, adding functionality, or modifying behavior. Explores intent, requirements, and design before implementation.

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

Update all README.md and CLAUDE.md files to reflect current codebase state. Combines update-docs and update-claudes. Also archives completed active plans.

Prepare context handoff when running low. Checks off completed todos in active doc, summarizes remaining work in Plan Mode, and prompts user to /clear.

Scan all CLAUDE.md files and agent_dev/README.md for `# Insights` sections. Present consolidated insights for human review.