DORA System – Repository & Technology Overview

The DORA platform is built as a set of modular repositories, aligned with the system architecture.
Each repository has a clearly defined responsibility, technology stack, and dependency surface.

DORA: Documentation Orchestration & Review Assistant

Tag line: Your intelligent partner in documentation.

---

1. cloudsteak/dora-mcp-server

Link: https://github.com/cloudsteak/dora-mcp-server

Purpose

Central MCP integration server exposing /mcp/... endpoints and routing tool-calls to internal agents.

Responsibilities

• Serve MCP capabilities, resources, tools, and tool-call endpoints
• Act as a router between the DORA backend and internal agents
• Hide agent implementation details behind a standardized MCP model

Dependencies

• DORA Backend (consumer of MCP)
• DORA Agents (internal services called by the MCP)

Technologies

• Python 3.13
• FastAPI
• Uvicorn
• httpx / aiohttp (for internal calls)
• Docker
• Helm chart (Kubernetes Deployment + ClusterIP)

---

2. cloudsteak/dora-backend

Link: https://github.com/cloudsteak/dora-backend

Purpose

Core intelligence layer of the platform, coordinating LLM interactions, RAG, and MCP client logic.

Responsibilities

• Communicate with LLM providers (OpenAI, Gemini, Azure OpenAI, etc.)
• Implement tool/function calling
• Call MCP server /mcp/... endpoints
• Integrate RAG (vector DB, document retrieval)
• Provide streaming API endpoint to the frontend
• Manage conversation state and agent orchestration

Dependencies

• MCP Server
• DORA Frontend
• Vector DB (optional for RAG)

Technologies

• Python 3.13
• FastAPI
• httpx (async streaming)
• LLM SDKs
• Redis / Qdrant (optional RAG)
• Docker
• Helm chart

---

3. cloudsteak/dora-frontend

Link: https://github.com/cloudsteak/dora-frontend

Purpose

The user-facing web-based chat interface.
Runs fully in Python using NiceGUI to avoid NodeJS build chains.

Responsibilities

• Provide a modern chat UI
• Send user messages to the backend
• Receive and display streamed responses
• Act as the ONLY publicly exposed component

Dependencies

• DORA Backend (REST / streaming API)

Technologies

• Python 3.x
• NiceGUI (Vue 3 + Tailwind internally, but Python-only development)
• httpx (streamed responses)
• Docker
• Helm chart
• Kubernetes Ingress for public access

---

4. cloudsteak/dora-agents

Link: https://github.com/cloudsteak/dora-agents

Purpose

Collection of all internal microservice “agent” implementations consumed by the MCP server.

Responsibilities

• Each folder contains an isolated agent microservice
• Provide specialized functionality (diagram generation, documentation writing, code analysis, etc.)
• Act as internal-only services reachable only via the MCP server

Dependencies

• MCP Server (caller)

Technologies

• Python 3.x
• FastAPI (typical for each agent)
• Specialized libraries (Mermaid renderer, Markdown tools, code parsers, etc.)
• Docker per agent
• Helm chart per agent
• ClusterIP networking only (no Ingress)

Example folder structure:

agents/ mermaid-agent/ documentation-agent/ code-analyzer-agent/

---

5. cloudsteak/dora-gitops

Link: https://github.com/cloudsteak/dora-gitops

Purpose

Central GitOps repository for deploying the entire DORA platform via ArgoCD.

Responsibilities

• Define ArgoCD Applications or ApplicationSets
• Reference Helm charts from each repository
• Maintain environment-specific values (dev, prod)
• Act as the declarative source of truth for production deployments

Dependencies

• ArgoCD
• All other DORA repositories

Technologies

• Kubernetes manifests
• ArgoCD
• Helm
• Kustomize (optional)

---

Summary Table

Repository	Purpose	Depends on	Technologies
dora-mcp-server	MCP central server	Backend, Agents	Python, FastAPI, Docker, Helm
dora-backend	LLM + RAG + MCP client	MCP server, Frontend	Python, FastAPI, httpx, Docker
dora-frontend	Chat UI	Backend	Python, NiceGUI, Docker
dora-agents	Internal agent microservices	MCP server	Python, FastAPI, Docker, Helm
dora-gitops	ArgoCD GitOps orchestration	All repos	ArgoCD, Helm, K8s