Toolkit for authoring and validating Claude Code skills and agents via a real script-backed check (description cap, PII, em-dash, Task* completeness, vague ask-user prose), not a self-scored rubric. Auto-invoke on 'create skill', 'validate skill', 'validate agent'.
5 skills · 4 agents
Idempotent macOS terminal installer for Ghostty, Oh My Zsh, Powerlevel10k, Glow, MesloLGS Nerd Font, plus optional markdown preview and clickable-path extras.
1 skill · hooks
Plan and run DIY builds with energy-aware work blocks, enforced breaks, and an inline child-safety veto gate. Project state lives in the build repo. Self-contained, Cowork-ready, no cross-plugin dependencies.
1 skill
Keep one canonical agent ecosystem usable on both Claude Code and OpenCode: generate OpenCode agents/commands/MCP from Claude sources, validate both runtimes' rules, detect drift, and ingest marketplaces or repos for OpenCode.
1 skill
Deterministic offline auditor for skills, agents, and prompts. Runs static checks against a directory of SKILL.md, AGENTS.md, or CLAUDE.md files and prints findings, critical then major then minor.
1 skill
Type-driven README updater. Detects repo family, confirms it with the user, asks for style, audience, and depth, then generates from the matching template workflow. Auto-invoke on 'update readme' and 'refresh readme'.
1 skill