pbakaus/impeccable-talks/starter
Claude Code and agent skill by pbakaus/impeccable-talks
- community
Checking that the SKILL.md file is still in the repository.
Demo skill for "The Dark Arts of Skill Engineering." A tiny, readable scaffold that shows five dark arts in miniature: routing (this file), a script that talks back (scripts/context.mjs), a hook that fights back (scripts/hook.mjs), making sub-agents argue (reference/argue.md), and forcing divergence (reference/diverge.md). Use when the user types /spellbook <command>, or asks to argue, diverge, or set up the demo.
Install
npx vskill@latest install pbakaus/impeccable-talks/starterSKILL.md
Spellbook (starter)
The workshop build: a pocket Impeccable you assemble as the talk climbs. It critiques and re-imagines sample/landing.html (a deliberately-median page) using five dark arts in miniature. It runs as-is; every art has a YOUR TURN marker where you extend it. Read it top to bottom, then make it yours.
Setup (always do this first)
DARK ART · the script talks back. Do not guess the project state. Run:
Follow whatever it prints. The script reads the environment (git branch, whether a NOTES.md exists) and returns a fresh, in-thread directive. That directive outranks anything written statically below, because it is shaped to the exact situation. If it prints NO_NOTES, create NOTES.md first. If it prints READY, continue.
Commands (the router)
DARK ART · route like a frontier model. This file is a thin router. Do not inline the technique here; load the one expert reference the command needs, and nothing else, so context stays cheap.
| Command | Load this, then follow it | |
|---|---|---|
/spellbook argue [page] | reference/argue.md | |
/spellbook diverge [page] | reference/diverge.md |
[page] defaults to sample/landing.html. If no command is given, ask which one the user wants. Never run both references at once.
The hook (already working, even when this skill is not invoked)
DARK ART · hooks that fight back. scripts/hook.mjs runs after every file edit (wire it up via .claude/settings.json, see README.md). It scans the touched file for a banned pattern and injects a system-reminder. You do not invoke it; it watches. Try editing a file to include TODO: fix later and watch it push back.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingdark-arts/starter/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/pbakaus/impeccable-talks/starter)