HN Showcase
Back to browse
GitHub Repository

Deterministic security guardrails for Claude Code. Six hooks that hard-block RCE, credential exfiltration, SSRF, and prompt injection before they execute. Plus an MCP server for every other agent.

5 starsShell

Deterministic security guardrails for Claude Code

by humblejedi·Apr 8, 2026·3 points·0 comments
Visit ProjectView on HN

AI Analysis

●●●BangerBig BrainSolve My Problem

Six shell hooks hard-block RCE and exfiltration before Claude Code executes anything.

Strengths
  • •PreToolUse hooks intercept commands before execution, not just logging after the fact
  • •MCP server extension makes guardrails available to any MCP-aware agent, not just Claude
  • •Self-aware security: installer warns against curl|bash and encourages reading the script first
Weaknesses
  • •Only works with Claude Code and MCP-aware agents, not general AI tool security
  • •Early project with just 2 stars and 1 fork—unproven in production environments
Category
Security
Target Audience

Developers using Claude Code and other MCP-aware AI agents

Similar Projects