abase-dcg

Block dangerous Git, filesystem, database, container, and cloud commands before execution.

Updated Feb 16, 2026

Search Tags:#git #safety #rust #ai safety #guard #command blocking

Ready to run

npx skills add https://github.com/mikegogulski/abase-django-blog-integration-test --skill abase-dcg

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: abase-dcg
Source: https://github.com/mikegogulski/abase-django-blog-integration-test/tree/main/.agents/skills/abase-dcg
Command: npx skills add https://github.com/mikegogulski/abase-django-blog-integration-test --skill abase-dcg

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill acts as a critical safety layer for AI coding agents, preventing accidental execution of destructive commands that could lead to data loss or system instability.

Core Features & Use Cases

Destructive Command Prevention: Intercepts and blocks dangerous commands related to Git, filesystem operations, and more before they run.
Whitelisting & Fail-Safe: Prioritizes safe commands and allows unrecognized ones by default, ensuring workflows aren't broken.
Modular Packs: Extensible system with packs for various domains like databases, containers, and cloud providers.
Use Case: An AI agent attempting to run rm -rf ./src will be blocked by DCG, with a clear explanation, preventing accidental deletion of source code.

Quick Start

Use the abase-dcg skill to protect your system from dangerous commands.

Frequently Asked Questions about abase-dcg

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I block dangerous git commands executed by AI agents?▼

You can block dangerous git commands by installing a safety hook that intercepts execution. This Skill provides a Rust-based guard that prevents destructive commands like force pushes or hard resets before they run.

How does command blocking prevent accidental data loss during AI coding?▼

Command blocking prevents accidental data loss by intercepting dangerous filesystem and git operations before execution. It uses a whitelist-first architecture to stop commands like `rm -rf` while allowing safe operations to proceed.

Can I use a command guard with Claude Code to intercept filesystem operations?▼

Yes, this Skill functions as a high-performance Rust hook specifically for Claude Code. It intercepts filesystem, database, container, and cloud provider commands to ensure agent safety.

Does command blocking introduce latency to AI coding workflows?▼

No, command blocking does not introduce noticeable latency because the hook uses SIMD-accelerated filtering. This ensures sub-millisecond processing times for evaluating commands.

What happens when a command guard encounters unrecognized commands?▼

When the command guard encounters unrecognized commands, it allows them to execute by default. This fail-safe whitelisting approach ensures your AI workflows are not broken by unknown but potentially safe commands.

Can I extend command blocking rules to cover database and cloud provider operations?▼

Yes, you can extend command blocking rules to cover databases and cloud providers using modular packs. This extensible system allows you to add specific safety domains as needed.