Azure DevOps RUP Planning Assistant
This project helps AI coding tools turn planning conversations into RUP-style SDLC artifacts and then persist them into Azure DevOps.
The planning vocabulary is RUP language:
- Stakeholder Request
- Functional Requirement
- Non-Functional Requirement
- UX Artifact
- Technical Requirement
- Architecture
- Technical Documentation
- Delivery Slice
- Task
Azure DevOps process terms such as CMMI Requirement, Scrum Product Backlog Item, Agile User Story, or Basic Issue are selected only at write time after the assistant reads the target project metadata.
Supported clients:
- Claude Code
- VS Code with GitHub Copilot
- OpenAI Codex
What It Does
| SDLC role | Output |
|---|
| Stakeholder Analyst | Stakeholder request, scope, business value, affected users, success measures |
| Requirements Analyst | Functional and non-functional requirements with acceptance criteria |
| UX Designer | User journeys, screen flows, Figma-ready screen specifications, accessibility notes, UX acceptance criteria |
| Solution Architect | Cohesive architecture views, boundaries, runtime flows, deployment, data, security, operations, technical requirements, ADR candidates |
| Technical Writer | Traceable technical documentation and Azure DevOps wiki-safe Mermaid diagrams from confirmed architecture |
| Delivery Planner | Delivery slices, estimates, dependency order, sprint or iteration recommendation |
| Implementation Lead | Task breakdown with sequencing and done criteria |
Nothing is created in Azure DevOps without confirmation.
Routes
Preferred routes:
| Route | Use it when |
|---|
/capture-request <description> | Capture a stakeholder request or change need |
/define-requirements <request> | Derive functional and non-functional requirements |
/design-ux <requirements> | Produce UX flow, screen specs, Figma guidance, and UX acceptance criteria |
/plan-requirement <description> | Refine one requirement |
/document-solution <requirements or architecture> | Produce technical documentation and diagrams |
/plan-delivery <requirement> | Produce delivery slices, estimates, and iteration placement |
/plan-task <delivery-slice> | Produce implementation tasks |
Natural planning intent should also trigger the workflow. For example, "I want to setup a highly secure home lab exposed through Cloudflare Tunnel" maps to /capture-request and starts the Stakeholder Analyst phase.
Development Readiness Gate
Development does not start until the request is traceable to an existing approved Azure DevOps work item or a confirmed RUP planning artifact. If the user asks for work that is not already represented in Azure DevOps, the assistant captures it as a new Stakeholder Request or Change Request and runs the SDLC workflow first.
User-facing work must include the UX Designer phase before development starts, unless UX is explicitly marked not applicable for non-UI work.
Architecture Quality
Architecture is not a technology list. It must define system boundary, actors, components, runtime flows, deployment, data, security, operations, decisions, tradeoffs, and open questions. Technical documentation must use the confirmed architecture and Azure DevOps wiki-safe Mermaid diagrams: ::: mermaid blocks and graph TD; or graph LR; for flowcharts.
Azure DevOps Mapping
Before writing work items, the assistant:
- Calls
mcp_ado_wit_list_backlogs.
- Reads target work item type metadata with
mcp_ado_wit_get_work_item_type.
- Chooses valid fields for estimates, acceptance criteria, requirement classification, tags, and iteration.
- Creates parent-before-child with
mcp_ado_wit_add_child_work_items.
- Updates only discovered fields with
mcp_ado_wit_update_work_item.
- Verifies traceability with
mcp_ado_wit_get_work_item.
Install
Pass the Azure DevOps organization during install. Pass the default project and optional team as well when you want a fully noninteractive setup.
Online install:
Windows PowerShell:
$env:ADO_MCP_ORG = "<your-azure-devops-org>"
$env:ADO_MCP_PROJECT = "<your-default-project>"
$env:ADO_MCP_TEAM = "<your-default-team>" # optional
$env:ADO_MCP_CLIENTS = "All" # optional: All, Claude, VSCode, Codex
$env:ADO_MCP_FORCE = "1" # optional: replace existing config blocks
irm https://tinyurl.com/yc3wvu6u | iex
Linux/macOS:
export ADO_MCP_ORG="<your-azure-devops-org>"
export ADO_MCP_PROJECT="<your-default-project>"
export ADO_MCP_TEAM="<your-default-team>" # optional
export ADO_MCP_CLIENTS="All" # optional: All, Claude, VSCode, Codex
export ADO_MCP_FORCE="1" # optional: replace existing config blocks
curl -fsSL https://tinyurl.com/bdfuef4w | bash
Repo-local install:
.\scripts\install.ps1 -Organization <your-azure-devops-org> -Project <your-default-project> -Team "<your-default-team>"
