Kaiza MCP Server

# START HERE: What is ATLAS-GATE-MCP? *This page explains ATLAS-GATE-MCP in plain English, with no technical jargon. It's designed for anyone encountering this project for the first time.* --- ## The Problem We Solve (In Plain English) Imagine you hire a very smart assistant (an AI model like ChatGPT) to help you with work. You give it powerful abilities—it can read files, write code, make decisions, launch processes. **The Challenge**: How do you make sure this assistant only does what you approved? How do you prevent mistakes? How do you prove to auditors and regulators what it actually did? **ATLAS-GATE-MCP is the answer.** It's like a security guard for AI assistants—it checks everything they want to do, ensures they follow the rules, and keeps a detailed record of their actions. --- ## What Does It Actually Do? ### **1. Controls What AI Can Do** - Decides which operations are allowed - Requires approval before important actions - Prevents accidental damage or misuse ### **2. Keeps A Record** - Logs everything the AI does - Can't be deleted or faked (tamper-proof) - Proves compliance to auditors ### **3. Enforces Security** - Prevents unauthorized access - Validates all input - Isolates the AI from direct system access ### **4. Works With Modern AI Tools** - Works with Claude, ChatGPT, Windsurf, and similar tools - Uses a standard protocol (MCP) that AI systems understand - No special training required for the AI --- ## Real-World Analogy Think of a bank: | Bank Element | ATLAS-GATE-MCP Equivalent | |---|---| | Customer wants to withdraw money | AI wants to modify a file | | Teller verifies customer ID | System validates the request | | Checks account has funds | Verifies authorization exists | | Records the transaction | Logs the operation | | Cannot delete old records | Audit log is immutable | | Auditor reviews records | Compliance team reviews logs | --- ## Who Should Use This? ### **YES, if you are:** - **A company** that uses AI tools to help with development - **Security-conscious** and need to prove governance - **Regulated** by GDPR, SOC 2, HIPAA, or ISO 27001 - **Enterprise-grade** requiring audit trails - **Risk-averse** and want controls on AI actions ### **MAYBE, if you are:** - **A startup** with limited compliance requirements - **Building experimental** projects where flexibility matters more - **A small team** where everyone knows everyone ### **NO, if you are:** - **Building a hobby project** alone - **Needing maximum speed** over security - **Not concerned with audit trails** --- ## How It Works (Simplified) ``` Your AI Assistant ↓ [Request: "Write this file"] ↓ ATLAS-GATE-MCP Security Gateway ├─ Is this approved? ✓ ├─ Is this allowed? ✓ ├─ Is the input safe? ✓ └─ [Log the action] ↓ [File written] ↓ [Audit log updated] ``` --- ## Key Concepts (Explained Simply) ### **Plan** A document that says "I approve this AI to do this work." Like a purchase order for AI actions. ### **Audit Trail** A detailed log of everything the AI did, when, and why. Like a receipts book that can't be edited. ### **Role** Different AI assistants have different permissions: - **Windsurf Role**: Can read and write (do work) - **Antigravity Role**: Can only read (analyze, plan, review) ### **Sandbox** A restricted environment where the AI can only interact through approved channels. Like a sandbox at the beach—the AI can play but can't escape. ### **MCP (Model Context Protocol)** The standard way AI assistants talk to tools. Like how your phone has standard charging ports so you can use different chargers. --- ## What Problems Does It Solve? | Problem | How ATLAS-GATE-MCP Helps | |---------|--------------------------| | "Did the AI follow the rules?" | Complete audit trail proves it | | "Can I trust an AI with code changes?" | Requires approval before changes | | "What if something goes wrong?" | Full log to trace what happened | | "How do I pass a security audit?" | Built-in compliance reporting | | "Can I prevent harmful actions?" | Content validation catches issues | | "Who did what and when?" | Cryptographically signed logs | --- ## Getting Started (Depending on Your Role) ### **I'm an Executive or Decision-Maker** → Read: [Enterprise Deployment Guide](./docs/enterprise-guide/DEPLOYMENT.md) and [Compliance Guide](./docs/enterprise-guide/COMPLIANCE.md) ### **I'm an IT/Ops Person** → Read: [Installation Guide](./docs/user-guide/INSTALLATION.md) and [Configuration Guide](./docs/user-guide/CONFIGURATION.md) ### **I'm a Developer** → Read: [Beginner's Guide](./docs/user-guide/BEGINNER_GUIDE.md) and [Development Setup](./docs/contributor-guide/DEVELOPMENT_SETUP.md) ### **I'm a Security Professional** → Read: [Security Model](./docs/architecture/SECURITY_MODEL.md) and [Audit Readiness](./docs/enterprise-guide/AUDIT_READINESS.md) ### **I Want to Contribute** → Read: [Contributing Guide](./docs/contributor-guide/CONTRIBUTING.md) --- ## The 2-Minute Version **ATLAS-GATE-MCP is:** - A security gateway for AI assistants - Ensures AI follows rules and gets approvals - Keeps tamper-proof records of what AI does - Built for enterprise compliance and security **Use it if you:** - Give AI systems access to critical work - Need to prove governance to auditors - Want to prevent or catch mistakes **It provides:** - Authorization (only approved actions) - Audit trails (complete records) - Compliance features (SOC 2, GDPR, ISO 27001) - Security controls (content validation, sandbox isolation) --- ## Glossary (Plain English Terms) | Term | Means | |------|-------| | **Audit Trail** | A record of what happened, when, and by whom (can't be changed) | | **Authorization** | Permission to do something (requires approval) | | **Compliance** | Following rules (legal, industry, or company) | | **Governance** | The rules and processes for making decisions | | **Sandbox** | A restricted area where things can run safely without affecting the rest of the system | | **Validation** | Checking that input is safe and correct | | **Tamper-proof** | Can't be secretly changed (you'd know if someone tried) | | **MCP** | Standard language AI assistants use to talk to tools | | **Windsurf** | An AI tool that can read and write | | **Antigravity** | An AI tool that can only read and analyze | --- ## Next Steps 1. **Understand the full picture** → [Documentation Hub](./docs/README.md) 2. **Install and try it** → [Installation Guide](./docs/user-guide/INSTALLATION.md) 3. **Read real examples** → [Use Cases](./docs/user-guide/EXAMPLES.md) 4. **Get technical details** → [Architecture](./docs/architecture/ARCHITECTURE.md) 5. **Contribute code** → [Contributing Guide](./docs/contributor-guide/CONTRIBUTING.md) --- ## Need Help? - **Question?** → Check [FAQ](./docs/user-guide/FAQ.md) - **Problem?** → See [Troubleshooting](./docs/user-guide/TROUBLESHOOTING.md) - **Unclear term?** → Look up [Glossary](./docs/GLOSSARY.md) - **Want to contribute?** → [Contributing Guide](./docs/contributor-guide/CONTRIBUTING.md) - **Security issue?** → Email security@atlas-gate-mcp.org --- **Version**: 2.0.0 **Last Updated**: February 2026 **Made for**: Everyone (technical or not)