Repolode
Submit
lennney

lennney/stop-that-shit

A multi-platform hook and skill guard for AI coding agents that stops unrequested hashes, checksums, and scope creep in Codex, Claude Code, OpenCode and other hosts.

View on GitHubWebsite
Stars
3,710
Forks
76
Open issues
4
Last commit
2 days ago

What is it?

Stop That Shit is a Skill and Guard pair for AI coding agents. The Skill gives the agent a decision order for judging whether a change is needed: it names four failure patterns (scope creep, unneeded hashing or hypothetical hardening, violating the user's stated intent, and repeating finished review or test work) and asks the agent to reuse existing capabilities before adding new mechanisms.

The Guard runs as a host hook before supported tool calls. Under a declared mode such as review, answer, or monitor, it blocks file writes, asks before adding dependencies, enforces a subagent budget, stops recognizable hash operations unless explicitly allowed, and rejects writes outside a declared file list. Mode and parameters are set with a leading directive such as `$stop-that-shit review --`.

The project supports Codex, Claude Code, OpenCode, Hermes Agent CLI, Pi Coding Agent and Oh My Pi, and includes an optional companion Skill for trimming defensive filler from written text.

Key features

  • 01Skill-based decision order for scope, existing capabilities, and defensive code.
  • 02Host hook guard that blocks file writes in review, answer, and monitor modes.
  • 03Parameters to lock file paths, allow dependencies, cap subagents, or permit hash operations.
  • 04Supports Codex, Claude Code, OpenCode, Hermes Agent CLI, Pi, and Oh My Pi.
  • 05Optional text-rewriting Skill for removing defensive filler from prose.

Use cases

  • Keeping a coding agent to a read-only code review that reports issues without editing files.
  • Preventing agents from adding unused checksums or compatibility layers to exported files.
  • Constraining a change task to a declared list of files and dependencies.
  • Trimming verbose, hedged wording from drafted documentation or articles.

Quick start

  1. 1

    Install the plugin in Claude Code: validate the plugin from the repo root

    claude plugin validate .
  2. 2

    Add the local repository as a plugin marketplace

    claude plugin marketplace add ./
  3. 3

    Install the stop-that-shit plugin

    claude plugin install stop-that-shit@stop-that-shit
  4. 4

    Restart Claude Code or run /reload-plugins, then use the skill with a review instruction

Things to know

  • Requires Node.js 18+ for most hosts, and Node.js 22.19+ for Pi.
  • Guard does not block actions until a mode such as review, answer, monitor, or change is explicitly used.

In collections

Claude Code tools, skills & pluginsOpen-source AI agent frameworks & appsAI coding assistants & agentsAI agent skill libraries

Topics

#ai-agents#ai-coding#guardrails#codex#claude-code#agent-skills#scope-control#developer-tools#codex-cli#codex-plugin

FAQ

What is stop-that-shit?
A multi-platform hook and skill guard for AI coding agents that stops unrequested hashes, checksums, and scope creep in Codex, Claude Code, OpenCode and other hosts. Stop That Shit is a Skill and Guard pair for AI coding agents. The Skill gives the agent a decision order for judging whether a change is needed: it names four failure patterns (scope creep, unneeded hashing or hypothetical hardening, violating the user's stated intent, and repeating finished review or test work) and asks the agent to reuse existing capabilities before adding new mechanisms.
Is stop-that-shit free to use? What license does it have?
stop-that-shit is open source under the MIT license. Check the license terms before using it in commercial products.
Is stop-that-shit still maintained?
Actively maintained. The last commit to stop-that-shit was on Nov 5, 2026.
What can I use stop-that-shit for?
Keeping a coding agent to a read-only code review that reports issues without editing files. Preventing agents from adding unused checksums or compatibility layers to exported files. Constraining a change task to a declared list of files and dependencies. Trimming verbose, hedged wording from drafted documentation or articles.
How do I install stop-that-shit?
Install the plugin in Claude Code: validate the plugin from the repo root: claude plugin validate . Add the local repository as a plugin marketplace: claude plugin marketplace add ./ Install the stop-that-shit plugin: claude plugin install stop-that-shit@stop-that-shit Restart Claude Code or run /reload-plugins, then use the skill with a review instruction
Where can I get stop-that-shit?
The source code and installation instructions are on GitHub: https://github.com/lennney/stop-that-shit

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects