neat-little-package

Traces code history via git blame and log to explain why code exists in its current form. Use when asked about code history, reasoning behind changes, or "when/why did this change".

Commit changes with smart messages. Supports full and partial commits. Use whenever committing changes or handling pre-commit hook failures.

Use the gh CLI for GitHub interactions, plus setup and pushing. Use when interacting with GitHub, setting up GitHub tools, or pushing to remote.

Universal polyglot linting and per-project config management. Use when you need to lint files, understand tool selection logic, invoke linting from commands/agents, or manage mr-sparkle config.

Resolves git merge conflicts by analyzing intent from both sides. Use when merge conflicts are detected or when git operations fail with conflict errors.

Create a pull request with Problem/Solution format. Can optionally commit first. Use when opening PRs or when branch changes are ready for review.

User's README style preferences — ultra-terse, action-focused, <50 lines. Use when writing or reviewing any README file.

Idioms and gotchas for managing dotfiles with chezmoi — source-state naming/attributes, when to use templates vs modify_ scripts vs run_ scripts, package installation, externals, secrets, and machine differences. Use when adding, editing, or structuring chezmoi-managed dotfiles, or writing chezmoi templates/scripts.

Empirically test whether mr-sparkle's permission-trigger blocking hook earns its keep — run its blocked patterns with the hook on, then with mr-sparkle disabled, and see which patterns actually trigger a Claude Code permission prompt without it. Use when deciding whether to keep, narrow, or drop block_unneeded_permission_triggers.sh + bash_guidance.sh.

Verification discipline for auditing this repo's components against CURRENT Claude / Claude Code reality and the user's preferences, plus the routine for keeping box-factory current. Find spec/guidance deltas, verify them empirically (don't trust the docs OR your own assumptions), push back on the user, record evidence. Use when auditing the repo, "make sure this is right", checking for drift, or updating box-factory.

How to build Agent Skills — the default format for all new skills. Covers Claude skills, the open AgentSkills spec, cross-platform gotchas, portability decisions, and graceful degradation. Use when creating or reviewing any skill.

How to build Claude Code-specific components — agents, hooks, plugins, marketplaces, MCP servers. User preferences, gotchas Claude gets wrong, and architecture decisions. Use when creating Claude-only components that need tool restrictions, hooks, or plugin infrastructure. For skills, use the agent-skills skill instead — skills default to the universal AgentSkills format.

Interpretive guidance for generating OpenSCAD code for Gridfinity desktop/drawer organizers. Provides pattern selection frameworks, dimensional constraints, and stacking system integration specific to the Gridfinity ecosystem. Use when generating OpenSCAD files for Gridfinity bins, baseplates, or accessories.

Decision framework for choosing between 3D-printed home organization systems (Gridfinity, OpenGrid, Neogrid, French Cleat, Underware, Deskware). Provides interpretive guidance on system selection based on item size, mounting surface, and use case. Use when users ask which organization system to recommend or need help choosing between modular storage options.

Interpretive guidance for generating OpenSCAD code for NeoGrid drawer divider connectors. This is a HYBRID system - print connectors, buy dividers (MDF, plywood, acrylic). Provides material selection guidance, connector type decisions, and base system integration. Use when generating OpenSCAD files for NeoGrid junction pieces.

Interpretive guidance for generating OpenSCAD code for OpenGrid/MultiConnect wall-mounted organizers. Provides pattern selection frameworks, mounting system integration, and dimensional constraints specific to this ecosystem. Use when generating OpenSCAD files for OpenGrid items.

Reference guide for generating Underware cable management channel OpenSCAD code. Provides channel type selection, parameter quick reference, and QuackWorks integration. Use when generating OpenSCAD files for Underware cable routing systems.

Gotchas and conventions for writing Python scripts with UV inline metadata (PEP 723). Use when creating or editing Python scripts that use uv run.

Check that required tools and env vars are configured for Home Assistant MCP

Guidance for Home Assistant configurations. Use when writing automations, dashboards, or service calls for HA.