How does this skill prevent accidental Git errors?

The skill strictly prohibits force-pushes, branch deletions, and direct commits to the main branch, requiring the AI to use feature branches and descriptive naming conventions.

What happens when the agent is uncertain about a task?

The protocol requires the agent to stop and ask for human clarification rather than proceeding with ambiguous or risky actions.

How does it manage project scope?

It prevents the AI from inventing requirements, assuming unstated needs, or making unrequested improvements that weren't part of the approved plan.

Can the AI agent bypass these safety rules?

No, any attempt to perform a forbidden action triggers an immediate STOP command, preventing the action and requiring human intervention.

Does it help with repository security?

Yes, it enforces safety by forbidding the hardcoding of secrets, the disabling of security analyzers, and the bypassing of failing tests.

Forbidden Actions & Safety Guardrails

Name: Forbidden Actions & Safety Guardrails
Author: ReneBossen

byReneBossen

開発者ツール

Enforces strict safety protocols and project guardrails to prevent destructive Git operations, scope creep, and unauthorized code modifications.

概要

The Forbidden Actions skill serves as a critical safety layer for Claude Code agents, establishing a clear boundary of prohibited behaviors to protect repository integrity. It mandates professional development standards by blocking direct commits to main branches, preventing destructive Git operations, and halting unauthorized architectural changes or scope expansion. This skill is essential for teams who want to leverage AI automation while ensuring the agent adheres to strict branch management policies, security best practices, and approved project plans without making unrequested 'improvements.'

主な機能

Enforces security protocols by prohibiting hardcoded secrets and the disabling of analyzers.
Blocks unauthorized scope expansion and 'while-at-it' refactoring to maintain PR focus.
0 GitHub stars
Prevents destructive Git operations like force-pushing, rebasing, or direct main branch commits.
Protects API contracts and project architecture from unapproved modifications.
Mandates immediate stops for human clarification when encountering ambiguity or blockers.

ユースケース

Onboarding AI agents to production repositories with strict branch protection and security requirements.
Maintaining project scope during feature development to prevent technical debt and 'feature creep'.
Ensuring AI-driven code modifications comply with existing architectural policies and package management rules.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add renebossen/claude-agents forbidden-actions

For use in Claude.ai and ChatGPT

Download Skill

GitHub

概要

主な機能

Enforces security protocols by prohibiting hardcoded secrets and the disabling of analyzers.
Blocks unauthorized scope expansion and 'while-at-it' refactoring to maintain PR focus.
0 GitHub stars
Prevents destructive Git operations like force-pushing, rebasing, or direct main branch commits.
Protects API contracts and project architecture from unapproved modifications.
Mandates immediate stops for human clarification when encountering ambiguity or blockers.

ユースケース

Onboarding AI agents to production repositories with strict branch protection and security requirements.
Maintaining project scope during feature development to prevent technical debt and 'feature creep'.
Ensuring AI-driven code modifications comply with existing architectural policies and package management rules.