# Overview The System Context diagram provides the highest-level view of the Metis Platform, showing its system boundary, primary users, external system interactions, and key responsibilities. # System Boundary **Metis Platform** is an AI-powered codebase intelligence system that bridges the gap between code reality and documentation intent by extracting dual-model knowledge and making it available to humans and AI agents. --- # Primary Users & Actors 1. **Developer Users** - Access Metis to understand codebase structure, navigate dependencies, and receive AI-assisted context 2. **Knowledge Managers** - Configure extraction rules, curate knowledge, manage integrations 3. **Organization Admins** - Manage users, permissions, billing, workspace configuration 4. **CodeVista Agent** - AI agent for code generation, receives tasks and context via MCP 5. **TestVista Agent** - AI agent for test generation, receives test criteria and coverage insights via MCP --- # External System Interactions # Git Repositories (Code & Documentation Source) - **GitHub, GitLab, Bitbucket, On-premise Git** - Source of code and documentation - **Connection:** OAuth 2.0 or token-based authentication - **Data Flow:** Repository clone/analysis on demand or scheduled; change detection for incremental updates - **Metis Role:** Read-only access to repository contents # Document Sources (Unstructured Documentation & Knowledge Storage) - **Azure SharePoint & Teams** - Enterprise document repositories with versioning, permissions, and collaborative metadata - **Cloud Storage (AWS S3, Azure Blob Storage, Google Cloud Storage)** - Unstructured document repositories - **Direct File Upload (ZIP, TAR, individual files)** - Users can upload code archives or document collections directly through web UI or API - **Connection:** OAuth 2.0, service principals, API tokens, managed credentials, or direct file system access - **Data Flow:** Document ingestion on demand or scheduled; metadata extraction; change detection; version tracking - **Metis Role:** Read-only access to documents; extract knowledge from unstructured content; maintain document relationships # CodeWiki Service (Metis's AI-Powered Living Documentation Engine) - **Metis CodeWiki** - Internal service that automatically generates and maintains living documentation for analyzed repositories - **Capabilities:**Auto-generates structured documentation after every code commit - Creates architecture diagrams, class diagrams, sequence diagrams automatically using code analysis - LLM-powered chat agent for repository Q&A using extracted knowledge - Hyperlinked documentation tied to actual code lines and line numbers - Self-hosted; works with both public and private/on-prem repositories - Continuously updated as code and documentation change **Connection:** Internal service within Metis platform; no external dependencies**Data Flow:** Knowledge Base feeds code/doc analysis to CodeWiki; CodeWiki generates living docs; docs feed back into Knowledge Base as premium documentation source**Metis Role:** Generate auto-documentation; maintain wiki freshness; provide Q&A interface; power task generation with visual diagrams # AI LLM Services (Knowledge Processing) - **OpenAI (GPT-4o/4.1), Google Gemini, Anthropic Claude, xAI Grok, Alibaba Qwen, Meta LLaMA, XMainframe, DeepSeek, Zhipu GLM, Moonshot Kimi, or customer-hosted LLMs** - Power semantic understanding and task generation - **Connection:** API integration with authentication - **Data Flow:** Extracted code/doc chunks sent for semantic analysis; structured results returned - **Metis Role:** Orchestrate LLM calls for knowledge extraction and task generation # CodeIQ (Code Graph Service) - **CodeIQ** - Internal/shared code intelligence service owned by a separate team - **Connection:** Internal gRPC/HTTP within the Metis deployment boundary - **Data Flow:** Knowledge Extraction Engine sends repository and commit identifiers; CodeIQ returns code entities and relationships as code graph snapshots - **Metis Role:** Treat CodeIQ as the single source of truth for code-level semantics and symbol graphs # MCP Clients (Agent Integration) - **CodeVista Agent** - Receives tasks and code context via MCP resources - **TestVista Agent** - Receives test planning context via MCP resources - **Connection:** MCP Server protocol (standard MCP implementation) - **Data Flow:** Agents query knowledge resources, receive context bundles, report execution feedback - **Metis Role:** Expose extracted knowledge as queryable MCP resources # Identity & Authentication Services - **SAML 2.0 / OAuth 2.0 Providers** - Enterprise SSO integration - **Connection:** Standard SAML/OAuth flows - **Data Flow:** User authentication tokens, identity attributes - **Metis Role:** Validate user identity and permissions # Data Storage (External) - **MinIO or S3-compatible object storage (self-hosted)** - Backup and multi-region file storage in cloud-agnostic way - **Email services** - Notifications and alerts - **Monitoring services** - Performance and error tracking --- # System Responsibilities 1. **Knowledge Extraction** - Analyze repositories and extract meaningful knowledge from code and documentation 2. **Knowledge Organization** - Store, version, and organize extracted knowledge with semantic indexing 3. **Task Generation** - Create structured, actionable coding tasks (implementation, refactors, fixes) and testing tasks (unit, integration, exploratory testing charters) for developers and AI agents 4. **Context Provision** - Generate relevant, role-aware context bundles (code, tests, docs, telemetry) tailored to coding and testing workflows, optimized for AI agent consumption 5. **Multi-Tenancy** - Maintain complete isolation between organizational data 6. **Integration** - Expose knowledge through MCP protocol for AI agent consumption 7. **Security & Compliance** - Manage authentication, authorization, encryption, and audit logging --- # System Context Diagram ` ` --- # Key Characteristics **Actors Outside System:** - Users interact through web interface, APIs, and SSO - Agents communicate via MCP protocol - External services provide data sources and computational resources **System Flows:** - Developers browse and search knowledge through UI - Admins manage organization and users - Agents query tasks and context via MCP - Platform analyzes repositories and enriches knowledge with LLMs - All identity validation through enterprise SSO **Deployment Options:** - Managed SaaS (multi-region cloud) - On-premise / Air-gapped (customer-managed infrastructure)