Vibekit icon

Vibekit

Vibekit is an open-source safety layer and security proxy built by Superagent Technologies that provides a secure sandbox environment for AI coding agents. It solves the critical problem of accidentally exposing API keys, secrets, and sensitive data when using coding agents like Claude Code, Gemini CLI, and OpenAI Codex.

The platform runs as a CLI wrapper that executes any coding agent in an isolated Docker container, providing complete isolation from your local machine. It automatically detects and redacts sensitive information before it leaves your system, including API keys, credentials, and personal data. The tool offers full observability into what your AI agents are doing - which files they access, what commands they execute, and what data they transmit.

Developed as a response to real security incidents where developers accidentally exposed API keys through their coding agents, Vibekit provides zero-latency protection without affecting performance. It supports all major coding agents and works entirely offline without requiring cloud dependencies or internet connectivity.

Key safety features include real-time data redaction on both input and output, prompt injection defense to block malicious instructions, AI threat intelligence that learns from attack patterns, and compliance-ready infrastructure meeting GDPR, HIPAA, PCI, and SOC 2 requirements out of the box.

The platform is MIT licensed and developed by the Y Combinator W24 backed team at Superagent Technologies, founded by Alan Zabihi and Ismail Pelaseyed. Their goal is to make AI coding agents safe and trustworthy for enterprise use while maintaining the developer experience that makes these tools valuable.

Vibekit

No discussions yet

Be the first to start a discussion about Vibekit

Demo Video for Vibekit

Developer

Y Combinator W24 backed company building infrastructure for AI coding agents. Founded by Alan Zabihi and Ismail Pelaseyed, Superagent T…read more

Pricing and Plans

(Open Source)

Free

Free
  • Full source code access
  • Docker container isolation
  • Secret redaction
  • Observability tools
  • Support for all major coding agents
  • Offline operation
  • Community support

System Requirements

Operating System
Linux, macOS, Windows
Memory (RAM)
2GB minimum
Processor
Any modern CPU
Disk Space
500MB for Docker containers

AI Capabilities

Coding agent safety and security
Real-time data redaction
Prompt injection detection
Secret detection and masking
Container-based isolation
AI threat intelligence
Security compliance automation
Observability and monitoring
Agent behavior analysis