Kaiza MCP Server

# ATLAS-GATE-MCP: Enterprise-Grade AI Security Gateway [](https://github.com/dylanmarriner/ATLAS-GATE-MCP/actions) [](https://opensource.org/licenses/ISC) [](https://nodejs.org/) [](https://github.com/dylanmarriner/ATLAS-GATE-MCP/releases) [](./SECURITY.md) [](./docs/enterprise-guide/COMPLIANCE.md) --- ## What is ATLAS-GATE-MCP? **ATLAS-GATE-MCP is a production-grade security gateway that controls, audits, and governs AI agent operations.** It solves a critical enterprise problem: **How do you safely empower AI assistants to help with important work while maintaining security, compliance, and control?** ### The Problem Modern AI tools (Claude, ChatGPT, Windsurf) are powerful but risky in enterprise contexts: - ❌ No way to enforce approval workflows - ❌ No audit trail of what they changed - ❌ No compliance evidence for auditors - ❌ Risk of accidental data exposure or breaking changes ### The Solution ATLAS-GATE-MCP is a security layer that: - ✅ **Controls**: Only approved operations execute - ✅ **Audits**: Every action logged with cryptographic proof - ✅ **Complies**: Built for SOC 2, ISO 27001, GDPR - ✅ **Governs**: Plan-based authorization model - ✅ **Isolates**: Process-level sandbox prevents abuse --- ## Quick Start (3 Minutes) ### **1. Install** ```bash git clone https://github.com/dylanmarriner/ATLAS-GATE-MCP.git cd ATLAS-GATE-MCP npm install ``` ### **2. Start** ```bash npm run start:windsurf ``` ### **3. Try It** - Use Claude, ChatGPT, or Windsurf IDE - It will use ATLAS-GATE-MCP to safely execute tasks - Check `audit-log.jsonl` to see what happened [Full getting started guide →](./docs/user-guide/BEGINNER_GUIDE.md) --- ## Key Features ### **Authorization & Governance** - **Plan-Based Approval**: AI operations require explicit authorization - **Role-Based Access**: Windsurf (read/write) vs Antigravity (read-only) - **Workspace Isolation**: Operations scoped to approved directories ### **Audit & Compliance** - **Immutable Audit Trails**: Cryptographically signed operation logs - **Non-Repudiation**: Proof of who did what, when - **SOC 2 Ready**: Controls for audit certification - **ISO 27001 Aligned**: Information security management ### **Security & Isolation** - **Process Sandbox**: MCP-only, no direct system access - **Content Validation**: Static analysis prevents malicious patterns - **Input Validation**: Schema-based request validation - **Session Isolation**: Independent permission contexts ### **Enterprise Features** - **Multi-Tenant Support**: Multiple workspaces with separate controls - **Compliance Frameworks**: GDPR, HIPAA, FedRAMP ready - **Disaster Recovery**: Audit logs enable forensics - **Monitoring Integration**: JSON logs for SIEM integration --- ## Documentation **Choose your path:** ### **New to this project?** → [START_HERE.md](./START_HERE.md) — 5-minute overview in plain English ### **Want to use it?** → [User Guide](./docs/README.md#for-end-users) — Installation, configuration, usage ### **Want to contribute?** → [Contributor Guide](./docs/contributor-guide/CONTRIBUTING.md) — Development setup, standards ### **Deploying to production?** → [Enterprise Guide](./docs/enterprise-guide/DEPLOYMENT.md) — Deployment, compliance, audit ### **Understanding the architecture?** → [Architecture Docs](./docs/architecture/ARCHITECTURE.md) — System design, security model --- ## Real-World Example ### **Scenario: Code Review with AI** **Before ATLAS-GATE-MCP:** ``` Developer: "Claude, review this code and make improvements" Claude: *Reviews and modifies files directly* Developer: "What changed? When? Did you follow our standards?" Claude: "I made changes but there's no record..." ``` **With ATLAS-GATE-MCP:** ``` 1. Developer creates PLAN: "Review and improve app.js" 2. Manager approves the plan 3. Claude uses ATLAS-GATE-MCP 4. Claude can only modify files in the plan 5. Every change logged: timestamp, user, file, diff 6. Audit log proves compliance to auditors ``` Result: **Full control, complete audit, zero surprises.** --- ## Who Should Use This ### **Yes, if you:** - Use AI tools to help with development or operations - Need to pass security audits (SOC 2, ISO 27001) - Handle regulated data (HIPAA, GDPR) - Want to prove governance to executives/auditors - Are risk-averse and want controls ### **Maybe, if you:** - Are a startup with limited compliance needs - Prioritize speed over governance - Have a small trusted team ### **No, if you:** - Are building a hobby project alone - Don't use AI for critical work - Have no compliance requirements --- ## Security Model ### **Five Trust Boundaries** ``` ┌─────────────────────────────────────────────────┐ │ AI Assistant (Claude, ChatGPT, Windsurf) │ └────────────────────┬────────────────────────────┘ │ [Boundary 1: MCP Protocol] ↓ ┌─────────────────────────────────────────────────┐ │ ATLAS-GATE-MCP Security Gateway │ │ ┌──────────────────────────────────────────┐ │ │ │ 1. Plan Validation (is this approved?) │ │ │ │ 2. Role Checking (Windsurf/Antigravity) │ │ │ │ 3. Content Scanning (any malicious code?)│ │ │ │ 4. Access Control (can reach this file?) │ │ │ │ 5. Audit Logging (record everything) │ │ │ └──────────────────────────────────────────┘ │ └────────────────────┬────────────────────────────┘ │ [Boundary 5: Audit Trail] ↓ ┌────────────────────────┐ │ Immutable Audit Log │ │ (Cryptographically │ │ signed, tamper-proof) │ └────────────────────────┘ ``` [See detailed security model →](./docs/architecture/SECURITY_MODEL.md) --- ## Compliance Status | Standard | Status | Documentation | |----------|--------|---| | **SOC 2 Type II** | 🟡 In Progress (90%) | [SOC 2 Readiness](./docs/enterprise-guide/COMPLIANCE.md#soc-2-type-ii) | | **ISO 27001** | 🟡 Ready (85%) | [ISO 27001 Guide](./docs/enterprise-guide/COMPLIANCE.md#iso-27001) | | **GDPR** | 🟢 Compliant (95%) | [GDPR Compliance](./docs/enterprise-guide/COMPLIANCE.md#gdpr) | | **HIPAA** | 🟡 Supportable (80%) | [HIPAA Guide](./docs/enterprise-guide/COMPLIANCE.md#hipaa) | | **NIST CSF** | 🟢 Aligned (90%) | [NIST Framework](./docs/enterprise-guide/COMPLIANCE.md#nist-cybersecurity-framework) | [Full compliance matrix →](./docs/enterprise-guide/COMPLIANCE.md) --- ## Getting Help ### **Questions About Features?** → [FAQ](./docs/user-guide/FAQ.md) ### **Something Not Working?** → [Troubleshooting](./docs/user-guide/TROUBLESHOOTING.md) ### **Technical Details?** → [Architecture](./docs/architecture/ARCHITECTURE.md) ### **Want to Contribute?** → [Contributing Guide](./docs/contributor-guide/CONTRIBUTING.md) ### **Security Issue?** → Email: security@atlas-gate-mcp.org --- ## Project Status - **Version**: 2.0.0 (MCP-Only Sandbox Enforcement) - **Stability**: Production-Ready - **Node.js**: 18.0.0+ - **License**: ISC (permissive open-source) - **Development**: Active - **Support**: Community + commercial options available [View releases →](https://github.com/dylanmarriner/ATLAS-GATE-MCP/releases) --- ## How It Works (High-Level) ### **1. Authorization Phase** ``` AI: "I want to write file.js" ATLAS-GATE-MCP: "Show me the plan" AI: "Here's plan-001" ATLAS-GATE-MCP: "✓ Plan is approved, proceed" ``` ### **2. Validation Phase** ``` ATLAS-GATE-MCP: "Let me scan the content" ATLAS-GATE-MCP: "✓ No malicious patterns found" ATLAS-GATE-MCP: "✓ Schema is valid" ``` ### **3. Execution Phase** ``` ATLAS-GATE-MCP: "✓ Operation allowed" File System: <file modified> ATLAS-GATE-MCP: "Logging operation..." ``` ### **4. Audit Phase** ``` Audit Log: { "timestamp": "2026-02-04T14:30:45Z", "operation": "file_write", "file": "file.js", "plan": "plan-001", "status": "SUCCESS", "signature": "RSA-SHA256:..." } ``` --- ## Architecture ATLAS-GATE-MCP implements a **multi-layered security architecture**: ``` AI Clients → MCP Protocol Layer → Security & Governance Layer ↓ Request Validation → Plan Authorization → Content Analysis ↓ Tool Registry → File Operations, Plan Mgmt, Audit Logging ↓ Immutable Audit Trail ← Non-repudiation & Compliance ``` [Full architecture diagram →](./docs/architecture/ARCHITECTURE.md) --- ## Contributing We welcome contributions from the community! ### **Ways to Contribute** - Report bugs - Suggest features - Improve documentation - Submit code improvements - Help with testing ### **Getting Started** 1. [Read CODE_OF_CONDUCT.md](./CODE_OF_CONDUCT.md) 2. [Follow Contributing Guide](./docs/contributor-guide/CONTRIBUTING.md) 3. [Check out issues](https://github.com/dylanmarriner/ATLAS-GATE-MCP/issues) --- ## License ATLAS-GATE-MCP is licensed under the **ISC License** (permissive, similar to MIT). This means you can: - ✅ Use it commercially - ✅ Modify it - ✅ Distribute it - ✅ Use it privately Just include the license in distributions. [Read full license →](./LICENSE) --- ## Support & Community - **GitHub Issues**: [Report bugs](https://github.com/dylanmarriner/ATLAS-GATE-MCP/issues) - **GitHub Discussions**: [Ask questions, discuss ideas](https://github.com/dylanmarriner/ATLAS-GATE-MCP/discussions) - **Email**: info@atlas-gate-mcp.org - **Security**: security@atlas-gate-mcp.org --- ## Related Resources - [MCP Protocol Spec](https://modelcontextprotocol.io/) - [Security Whitepaper](./docs/enterprise-guide/SECURITY_CONTROLS.md) - [Compliance Framework](./docs/enterprise-guide/COMPLIANCE.md) - [Architecture Decision Records](./docs/adr/) - [Changelog](./docs/changelog/CHANGELOG.md) --- ## Acknowledgments ATLAS-GATE-MCP was built with inspiration from: - CNCF projects (Kubernetes, Prometheus, etc.) - Modern security frameworks (NIST, SOC 2) - Enterprise governance best practices - Open-source community standards Special thanks to all contributors and the community. --- ## Roadmap ### **2026 (Current)** - ✅ v2.0: MCP-Only Sandbox Enforcement - 🟡 SOC 2 Type II certification - 🟡 ISO 27001 alignment ### **2027** - Database integration for enterprise deployments - Advanced policy engine - Multi-server deployment support - HIPAA BAA and PCI-DSS support [Full roadmap →](./docs/maintainer-guide/ROADMAP.md) --- ## FAQ **Q: Do I need to understand MCP to use this?** A: No. ATLAS-GATE-MCP works transparently with your AI tools. You just use them normally. **Q: Can I use this in production?** A: Yes. v2.0 is production-ready with comprehensive audit logging. **Q: Will this slow down my AI assistant?** A: Minimal overhead (typically <100ms per operation). Built with performance in mind. **Q: Can I use this offline?** A: Yes. All operations are local; no external services required. **Q: Is it open-source?** A: Yes. ISC License (permissive, similar to MIT). [More FAQ →](./docs/user-guide/FAQ.md) --- **Latest Update**: February 2026 **Version**: 2.0.0 **Status**: Production-Ready **Learn More**: [Docs Hub](./docs/README.md) | [START_HERE](./START_HERE.md)