Metadata-Version: 2.4
Name: hardstop
Version: 0.0.1
Summary: The mechanical brake for AI-generated commands. Pre-execution safety validation for Claude Code.
Author-email: frmoretto <francesco.marinoni.moretto@gmail.com>
License-Expression: CC-BY-4.0
Project-URL: Homepage, https://github.com/frmoretto/hardstop
Project-URL: Repository, https://github.com/frmoretto/hardstop
Project-URL: Issues, https://github.com/frmoretto/hardstop/issues
Keywords: hardstop,ai-safety,command-validation,security,claude-code,mcp,pre-execution,fail-closed,llm,bash,shell
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Security
Classifier: Topic :: Software Development :: Quality Assurance
Requires-Python: >=3.9
Description-Content-Type: text/markdown

# 🛑 Hardstop

> The mechanical brake for AI-generated commands

**Status:** Placeholder - Full release coming soon

## What is Hardstop?

Hardstop is a pre-execution safety system for AI-generated shell commands. It acts as a fail-closed verification layer, blocking dangerous patterns (like `rm -rf ~/`, reverse shells, credential exfiltration) before they execute.

**Two-layer defense:**
- **Pattern matching** — Instant regex-based detection
- **LLM analysis** — Semantic analysis for edge cases

## Key Features

- **Fail-closed design** — If safety check fails, command is blocked (not allowed)
- **Cross-platform** — Unix + Windows patterns
- **Command chaining** — Analyzes all parts of piped/chained commands
- **Audit logging** — All decisions logged

## Coming Soon

- Full Claude Code plugin release
- MCP server integration
- Configurable security policies

## Links

- [GitHub Repository](https://github.com/frmoretto/hardstop)

## License

CC-BY-4.0
