AI for Developers
Agent skill (SKILL.md) Open source MIT 3.7k

stop-that-shit

A hook and skill guard that blocks unrequested hashes, checksums and scope creep.

Install

Install
$ claude plugin marketplace add ./ claude plugin install stop-that-shit@stop-that-shit
Or with the skills CLI
$ npx skills add lennney/stop-that-shit -a claude-code
For every project
$ npx skills add lennney/stop-that-shit -a claude-code -g

Or copy the skill folder into .claude/skills/ (project) or ~/.claude/skills/ (all projects).

For this project
$ npx skills add lennney/stop-that-shit -a codex
For every project
$ npx skills add lennney/stop-that-shit -a codex -g

Or copy the skill folder into .agents/skills/ (project) or ~/.agents/skills/ (all projects).

For this project
$ npx skills add lennney/stop-that-shit -a opencode
For every project
$ npx skills add lennney/stop-that-shit -a opencode -g

Or copy the skill folder into .agents/skills/ (project) or ~/.config/opencode/skills/ (all projects).

For this project
$ npx skills add lennney/stop-that-shit -a hermes-agent
For every project
$ npx skills add lennney/stop-that-shit -a hermes-agent -g

Or copy the skill folder into .hermes/skills/ (project) or ~/.hermes/skills/ (all projects).

Run claude plugin marketplace add ./ then claude plugin install stop-that-shit@stop-that-shit from a clone.

When to use

Install it if your agent keeps expanding the task, such as editing when you asked for a review or adding defensive code nobody asked for.

Stop That Shit is a skill, a guard and a check, called STSS, that keeps an agent on the task you gave it. The README names four habits, spelling out SHIT. Scope creep means fixing one bug and refactoring the module next door. Hashing and hypothetical hardening means adding checksums nobody reads. Intent violation means editing files during a review. Task thrashing means re-reading and re-testing what was already checked.

The author says piling rules into AGENTS.md just made that file bloat, so this moves the discipline into a hook and a skill. It isn't against thoroughness. Needed callers, migrations and tests still get done. It works with Codex, Claude Code, OpenCode, Hermes Agent and Pi. The README is in Chinese, with English and Korean versions linked.

Related skills

All skills